The internal/external frame of reference model in math, verbal, and general academic self-concept structure of elementary school children in Taiwan

This study examined math and verbal self-concept structure from Marsh's internal/external (I/E) frame of reference model using cognitive dimensions. Gender differences in the structure of academic self-concept were also examined. Math and verbal self-concepts on a cognitive level were defined as success expectations on certain familiar tasks. Two assumptions tested in the I/E model were: (a) Math and verbal self-concepts were uncorrelated, and (b) math achievement had a negative direct effect on verbal self-concept and verbal achievement had a negative direct effect on math self-concept.;The study was conducted using Academic Self-Esteem Scale, Cognitive Self-Concept Scales, and Achievement Tests in math and verbal areas for data collection. Subjects were 539 sixth-grade students (279 boys and 260 girls) in Taichung City in Taiwan during the year of 1991.;The I/E model assumptions were examined through path analysis for the latent variables. No strong support was found for the I/E model. Math and verbal self-concepts were highly correlated. Two equivalent models, either one with the path from verbal self-concept to math self-concept opened, or the other one with the path from math self-concept to verbal self-concept opened, were identified to fit the data better than the model with the path between the two self-concepts blocked. The significant negative direct effects of verbal achievement on math self-concept and of math achievement on verbal self-concept were not both found in each of the two equivalent models for boys, girls, and total subjects. The results differed from previous research measuring evaluative dimensions of math and verbal self-concepts, but supported the findings using cognitive dimensions in general. The results were explained by the differences of school education, of academic self-concept dimensions, of cultural backgrounds, or of physical and psychological development. The structure of self-concept differed very little for boys and girls, suggesting that no sex differences in cognitive math and verbal self-concepts were larger than what could be explained by corresponding sex differences in achievement. The influence of traditional stereotypes that boys are said to have better math ability than girls and girls are said to have better verbal ability than boys was not supported. The limitations and suggestions of this study were discussed.