Tests of the internal/external frames of reference model with subject-specific academic self-efficacy and frame-specific academic self-concepts.

H. W. Marsh's (1986) internal/external (I/E) frames of reference model posits that students* self-concepts result from simultaneous comparison of their competence to their peers' ability and their ability in other areas. The VE model failed to receive clear support with (a) subject-specific self-efficacy and (b) frame-specific self-concepts. Frame-specific selfconcepts were assessed by asking students to report their self-perceived capability in direct reference to the internal and external comparison frames. Contrary to the I/E model's assumptions, students' verbal and math self-concepts based on the internal comparison were positively correlated; achievement in one area negatively influenced both internal and external comparison-based self-concepts in the other area; and verbal and math self-concepts were positively correlated. Interestingly, most of the model's predictions were confirmed when the classical 1/E model structure was replicated, demonstrating that the hypothesized relations among self and achievement factors can be achieved without satisfying the model's theoretical provisions. The purpose of the present investigation was twofold. One was to test predictions of the internal/external (1/E) frames of reference model (Marsh, 1986) with subjectspecific academic self-efficacy to help illuminate the difference between academic self-concept and self-efficacy formation. The other was to test underlying assumptions of the I/E model with frame-specific academic self-concepts. Specifically, the present study examined whether students* verbal and math self-concepts were indeed positively correlated after the external comparison and negatively correlated after the internal comparison as the I/E model predicates.