According to the internal/external frame of reference model, academic achievement has a strong impact on people’s self-concept, both within and between subjects. We conducted a series of meta-analyses of k = 505 data sets containing the six bivariate correlations between achievement and self-concept in two subjects. Negative paths from achievement to noncorresponding self-concept, indicating dimensional comparison effects, were strongest when the subjects were dissimilar with regard to the math-verbal continuum, reduced but still significantly negative when both subjects belonged to the verbal domain, and near-zero when both subjects belonged to the math/science domain. Additionally, we found stronger positive paths from achievements to corresponding self-concepts, indicating social comparison effects, and stronger dimensional comparison effects for grades than for standardized test scores, and for older rather than younger students. We extend dimensional comparison theory by discussing these results with particular regard to the nonexistence of assimilation effects, the effects of subject similarity on dimensional comparison effects, and other moderators of dimensional comparison effects.

Students evaluate their domain-specific abilities by comparing their own achievement in a certain domain with the achievement of others (social comparison), with their own prior achievement (temporal comparison), and with their own achievement in other domains (dimensional comparison). This article is the first to analyze the simultaneous effects of social, temporal, and dimensional comparisons on students' academic self-concepts of various domains in experimental and field studies. In Study 1 (N ϭ 120), students judged their ability self-concept after having received experimentally manipulated social, temporal, and dimensional comparison feedback. In Study 2 (N ϭ 924), students had to rate their math and German self-concept and were asked to directly compare their achievement to social, temporal, and dimensional comparison standards. In the longitudinal Study 3a (N ϭ 3,054) and 3b (N ϭ 14,008), the three types of comparisons were modeled in an extended internal/external frame of reference model (Marsh, 1986) containing paths from math and verbal achievement level and achievement change to math and verbal self-concept. In all studies, social, temporal, and dimensional comparisons showed significant effects on self-concept. For each comparison process, downward comparisons with worse-off standards resulted in higher self-concepts, whereas upward comparisons with better-off standards resulted in lower self-concepts. These results are in accordance with the theories underlying social, temporal, and dimensional comparison processes and support their integration into a combined framework. Educational Impact and Implications StatementAcademic self-concepts describe students' subjective beliefs concerning their strengths and weaknesses in different school subjects. A high self-concept is of major importance; for example, students tend to show better achievements in a certain subject if they have a high self-concept in the respective subject. This study shows that three comparison processes bear an influence on the formation of students' academic self-concepts: social comparisons, temporal comparisons, and dimensional comparisons. Students develop a higher academic self-concept if they judge their achievement superior in comparison with their classmates (social comparison), with their prior achievement (temporal comparison), and with their achievement in other subjects (dimensional comparison). Teachers can make use of these findings by stressing specific comparison processes to support their students in the development of positive academic self-concepts.

The 2I/E model is an extension of Marsh’s (1986) classical internal/external frame of reference (I/E) model. Whereas the classical I/E model describes the formation of academic self-concept by the joint operation of social (external) and dimensional (internal) comparisons, the 2I/E model integrates temporal comparisons as an additional internal comparison process. In the 2I/E model, paths from achievement levels to self-concepts in the same subject are assumed to represent social comparisons, paths from achievement levels to self-concepts in another subject are assumed to represent dimensional comparisons, and paths from achievement changes to self-concepts in the same subject are assumed to represent temporal comparisons. The present research validated the 2I/E model in six empirical studies with a total of more than 20,000 students. Generally, there were strong positive social comparison paths, moderate negative dimensional comparison paths, and small positive temporal comparison paths. Furthermore, the relations of the 2I/E model were replicated with both grades and standardized test scores as achievement indicators (Study 1 and Study 2). Dimensional comparison effects were stronger between dissimilar subjects, like math and German, than between similar subjects, like German and English (Study 2 and Study 3). Also, the relations of the 2I/E model were shown when prior self-concepts were controlled for (Study 4). Overall, the 2I/E model seems to have the potential to enrich future theory and research on self-concept formation and comparison processes.

No abstract

No abstract