When Mathematical Learning Experiences Match Self-Efficacy Beliefs: Content Specificity and Cognitive Performance Across 49 Education Systems

Self-efficacy is task-referenced, yet it is often treated as a single subject-level belief. Drawing on the specificity and correspondence principles of self-efficacy theory, this study asks whether learning experiences that more closely match a judged capability show stronger and more generalizable relations with content-specific mathematics self-efficacy. PISA 2022 data from 173,641 adolescents in 49 education systems were analyzed. Five learning experiences ranged from direct task exposure to process-focused cognitive activation and context-mediated digital inquiry. Item-level analyses supported the intended within-domain distinctions, while exact loading comparability was less consistent for task exposure. System-specific weighted structural models accounted for school clustering and all ten mathematics plausible values, and the resulting estimates were synthesized with random-effects meta-analysis. Direct task exposure produced the clearest pattern. Formal/applied exposure was more strongly associated with formal/applied self-efficacy, whereas reasoning/21st-century exposure was more strongly associated with reasoning-oriented self-efficacy. The combined task-matching summary was positive in all 49 systems, and its random-effects prediction interval remained above zero. Cognitive activation showed differentiated but nonliteral correspondence, while digital inquiry was positively associated with both beliefs and varied more across systems. Both self-efficacy dimensions were positively associated with mathematics performance when examined separately, although their unique criterion relations differed in joint models. The findings show that mathematics self-efficacy has meaningful within-domain structure and that experience–belief correspondence depends on how closely learning experiences instantiate the capability being judged.