Linear and nonlinear relationships between self-evaluation and self-evaluation bias with grades

There is an ongoing debate whether a bias in self-evaluated competence (SE bias) affects academic achievement, typically assessed as grades. Different rationales about possible relations (e.g., linear, nonlinear, or no relations) between SE bias and academic achievement have been proposed. We compared these rationales in a study with 504 secondary school students in highest academic track secondary schools in math, using response surface analysis (RSA). At t1, students' self-evaluated math competence, their competence in math (i.e., objective competence test scores), and their academic achievement in math (i.e., math grades) were assessed. At t2, stu-dents' math grades were assessed again. The model that fit the data best was the beneficial self-evaluation and competence model, positing only linear relationships of self-concept and competence with change in grades beyond control variables (i.e., gender, parental education, number of books at home) and no SE bias effects.