The Relationship Between Academic Self-Concept and Academic Achievement in College Students

Abstract

Academic self-concept is considered central to the identity and development of college students. Previous studies of academic self-concept of college students have largely relied on only two time points to study changes that occur during college. Additionally, prior studies have found a positive association between academic self-concept and academic achievement, but have employed data techniques that limit the conclusions drawn to correlational associations. This study investigated the changes in academic self-concept over the four years of undergraduate education using a large but non-random sample of students at a highly selective institution. Additionally, using auto-regressive cross-lagged models, it explored the causal ordering of academic self-concept and academic achievement. Finally, the study explored whether changes in academic self-concept, and its relationship to academic achievement, were different for men and women. The results suggested an initial decline in academic self-concept during the first year of college, consistent with the Big-Fish-Little-Pond-Effect, followed by an increase in academic self-concept at graduation. The study found men and women experienced similar patterns of change in academic self-concept through the college years, but that women’s academic self-concept was consistently lower than men’s academic self-concept. For men, higher academic achievement was followed by higher academic self-concept at all time points. Among women, however, there was evidence of reciprocal effects between academic self-concept and academic achievement, especially in the first year of college. These findings indicate that lower academic self-concept in women could be academically disadvantaging them compared to men, particularly early in their collegiate experience. The study also examines the measurement properties of CIRPs Academic Self-Concept scale and the adequacy of auto-regressive cross-lagged models in place of multi-level auto-regressive cross-lagged models. Policy and research implications, as well as future directions for research, are discussed.