Long-term associations between self-concept and achievement : Examining the reciprocal internal/external frame of reference model from a developmental perspective

Abstract

Theoretical background:The idea of a good academic self-concept as an exclusively positive characteristic of students has been questioned both on a theoretical and on an empirical level during the last decades. While research on the reciprocal effects model (REM) showed positive reciprocal effects between self-concept and achievement within the same domain (i.e., skill development and self-enhancement effects), research on the internal/external frame of reference model (I/E) revealed negative contrast effects between self-concept and achievement across different domains (Huang, 2011). Möller et al. (2011) combined the REM and the I/E models by longitudinally examining the interplay between self-concept and achievement in multiple domains. Studies examining the reciprocal internal/external frame of reference model (RI/E) were in support of its assumptions regarding skill development, self-enhancement, and negative contrast effects (e.g., Niepel, Brunner, & Preckel, 2014). It is important to note, that studies in this field have focused on associations between adjacent measurement occasions. However, as the theoretical assumptions from the RI/E model refer to development in the long-term, it is important to also model and to examine the reciprocal development of self-concept and achievement in different domains over an extended time period such as high school.Research question:The aim of the present study was to replicate previous findings and to extend them by taking a long-term perspective on the RI/E model and to test the hypotheses of skill development, self-enhancement, and contrast effects across high school using latent growth modeling (LGM). The RI/E model was examined for mathematics, German, and English.Method:916, 719, and 647 students participated in the first (2012), second (2013), and third (2014) wave of assessment, respectively (mean age at T1 = 15.6 years; 56 % female). Students’ academic self-concepts in mathematics, German, and English were assessed using the Self-Description Questionnaire (Marsh & O’Neill, 1984). Mid-term grades were obtained from the school administrations. A multivariate cross-lagged model (CLM) was modeled to replicate previous findings. Additionally, two multivariate parallel process LGMs were modeled to examine the hypotheses of skill development, self-enhancement, and contrast effects across high school.Results:Theoretical assumptions as postulated by the RI/E model were mostly supported in analyses focusing on adjacent measurement occasions (CLM). Additionally, new insights were gained from analyses focusing on the entire time frame of high school (GLMs). Within mathematics, we found negative associations between the initial scores of achievement and the development of self-concept, as well as negative associations between the initial scores of self-concept and the development of achievement. Further, the development of achievement and self-concept were found to be positively correlated. The same pattern was found for English. As for German, the initial score of achievement was found to be negatively associated with the development of self-concept. However, as achievement in German was found to be highly stable over time and to have a non-significant variance, no associations with the development of achievement in German were found both within and between domains. Regarding associations between mathematics and English, initial scores of achievement of each domain were not found to be associated with the development of self-concept in the other domain and vice versa. Additionally, the development of self-concept in mathematics was not found to be associated with the development of achievement in English, and the development of self-concept in English was not found to be associated with the development of achievement in mathematics. In sum, evidence for negative self-enhancement and skill development effects was found, while no evidence was found for contrast effects. Differences and commonalities between results stemming from the two analysis strategies are going to be discussed with respect to the theoretical framework of the RI/E model.