Examining grades, achievement test scores, and three intelligence facets within an extended I/E model

Domain-specific academic self-concepts (ASCs) are mental representations of one’s abilities that are specific to a particular school subject. According to the internal/external frame of reference (I/E) model, achievement influences the formation of ASCs via social and dimensional comparison processes: within one domain students not only compare their achievement with their peers (social comparison) but also with their own achievements in other domains (dimensional comparison). However, students’ intelligence as an indicator of achievement has been neglected within the I/E model framework. Yet, intelligence is known to be an important determinant of academic success and thus it can be assumed to influence the formation of academic self-concepts, too. Hence, this research examined the links between verbal, numerical, figural intelligence and domain-specific ASCs while controlling for grades and achievement test scores. We drew on N = 382 German students to analyze verbal, numerical, and figural intelligence, German and math achievement, domain-specific self-reported grades and ASCs in math, physics, German, and English. We performed structural equation modeling using Mplus 8 with grades, achievement scores and intelligence facets specified as manifest predictors and domain-specific ASCs as latent criteria. Positive within-domain relations indicating social comparison effects were found between math, physics, German, and English grade to their corresponding self-concept as well as from math achievement to math ASC. Dimensional comparison effects are implied by a positive cross-domain path between physics grade and math ASC and by negative cross-domain paths from math grade to German, physics, and English ASC. Further, a positive cross-domain relation was found between math achievement and physics ASC while a negative cross-domain path was found between German achievement and math ASC. With regard to the intelligence facets, positive paths were found between numerical intelligence and physics ASC as well as between verbal intelligence and English ASC. Our findings thus suggest numerical and verbal intelligence to be valid predictors of ASC formation in an extended I/E model beyond grades and test scores. Overall, the proposed I/E model permitted meaningful relations to be drawn between domain-specific achievement indicators and ASCs suggesting that these variables provide incremental validity.