The relationship between cognitive activation and math anxiety in Japan: moderating roles of socioeconomic status and math self-efficacy

Math anxiety undermines students’ motivation and achievement and is a pervasive challenge in mathematics education. Although prior studies have shown that cognitively demanding instruction is associated with lower anxiety, the possibility of nonlinear effects and contextual moderators remains underexplored. Drawing on nationally representative data from 5,760 Japanese 15-year-olds in PISA 2022, this study examined the curvilinear relationship between cognitive activation (mathematical-thinking activation and reasoning-focused activation) and math anxiety, and the moderating roles of student SES, school SES, and math self-efficacy. Multilevel regression analysis revealed a plateau effect for mathematical-thinking activation: it reduced math anxiety at lower-to-moderate levels but showed diminishing returns at higher levels. Importantly, this pattern was most pronounced among low-SES students, consistent with a bounded compensatory interpretation, whereas high-SES students showed relatively stable anxiety levels. In contrast, math self-efficacy, although strongly predictive of anxiety as a main effect, did not moderate the relationship. These findings refine control-value and cognitive load theories by highlighting nonlinear and SES-dependent dynamics, and they underscore that instructional quality is associated with not only cognitive but also emotional outcomes. Practically, the results suggest the value of promoting mathematical-thinking activation in teacher training to potentially mitigate math anxiety, particularly among socioeconomically disadvantaged students.