Subject-specific interests and subject-specific self-concepts

Abstract

Abstract. Academic interest and academic self-concept are important correlated motivational variables. Their common factorial structure and their shared (confounded) and non-shared (unique) power for predicting subject-tied academic achievement has not yet been satisfactorily investigated. This study investigated 588 Chinese adolescents. Two subject-tied interests and the corresponding subject-tied self-concepts were measured. The associated school marks were collected. The major subject German (first foreign language) was chosen to represent the verbal oriented school subjects, the major subject Maths was chosen to represent the numerical oriented ones. Confirmatory factor analyses (CFAs) evidenced two motivational factors (academic interest and academic self-concept) within each school subject and two subject factors (German and Maths) within each motivational construct. Joined CFAs (which included all interest as well as all self-concept items) revealed four factors (interest German, interest Maths, self-concept German, self-concept Maths). Accordingly, four motivational scales were formed. Each scale consisted of five items. The scales had high reliabilities and displayed good convergent-divergent validities. Both motivational scales together accounted for 41.0% (German) and 22.3% (Maths) of the total achievement variance. Of these, the common (fused) shares were 19.8% (German) and 12.4% (Maths). The proportions of the total achievement variance that could be explained exclusively by only one of the two motivational variables amounted to 0.3% (interest) and 20.9% (self-concept) for German. Theses figures were 0.1% (interest) and 9.8% (self-concept) for Maths. The discussion underscores that subject-specific academic interest and subject-specific academic selfconcept, while highly correlated, should still be considered distinct constructs. A subject-tied self-concept remains an important predictor of academic achievement even after controlling for the assigned interest. The situation is different in the case of subject-tied interest. Its predictive power for academic achievement collapses after controlling for the associated selfconcept.