Replicating and extending the GI/E model: Social and dimensional comparison effects of achievement on test anxiety in math, physics, German, and English

The internal/external frame of reference model (I/E model; Marsh, 1986) depicts the formation of academic self-concept by assuming an interplay of social and dimensional comparison processes within and across different domains. Associations between achievement and self-concept are expected to be positive within one given domain due to social comparisons (i.e., external frame of reference) and negative across different domains due to dimensional comparisons (i.e., internal frame of reference). The Generalized I/E model (GI/E model; Möller, Müller-Kalthoff, Helm, Nagy, & Marsh, 2015) allows for the inclusion of other outcome variables besides academic self-concept, for example, interest or motivation. The present research aimed at applying the GI/E model to the construct of test anxiety, thereby replicating and significantly extending some first supportive findings on the validity of the GI/E model with test anxiety (Arens, Becker, & Möller, 2017). To this end, we expanded the scope of the GI/E model with test anxiety for the first time to four domains; namely math, physics, German, and English. For this purpose, we drew on a sample of N = 305 9th and 10th graders attending the highest ability school track (i.e., Gymnasium) from six different secondary schools in four German federal states. Analyses were carried out using structural equation modelling in Mplus 8. We used school grades in the respective domains as achievement indicators. Test anxiety was assessed separately for each domain while simultaneously differentiating between two components of test anxiety; namely, worry and emotionality. Our results suggested negative within-domain relations between achievement and test anxiety in all four domains (i.e., higher grades were associated with less test anxiety). The pattern of results for the relations across domains was not as clear. Positive links between achievement and test anxiety across domains (i.e., higher grades in domain A were associated with higher test anxiety in domain B) were weak and could only be established for the association between German achievement and students’ worry in mathematics. These findings partially supported the assumptions made by the GI/E model. As such, we replicated preexisting findings and added novel insights to the GI/E model. Our results were discussed within the framework of self-concept theory and research as well as dimensional comparison theory.