The relationship between academic self-efficacy and test anxiety: a meta-analysis of Chinese secondary school...

Existing empirical studies report inconsistent correlation coefficients, ranging from -0.105 to -0.792, for the association between academic self-efficacy and test anxiety among Chinese secondary school students. Although evidence indicates that academic self-efficacy predicts test anxiety more strongly than domain-general self-efficacy (global confidence across all life domains), most prior meta-analyses have focused on the latter rather than on academic self-efficacy (perceived competence in specific academic domains). Furthermore, few quantitative meta-analyses have systematically decomposed moderators contributing to cross-study heterogeneity, leaving a notable research gap. Literature was retrieved from CNKI and Web of Science. The present meta-analysis included 30 studies (N = 27,865) of Chinese secondary school students. Publication bias was evaluated via multiple complementary methods, including Egger's linear regression test, Duval and Tweedie's trim-and-fill, classic fail-safe N, and Begg and Mazumdar's rank correlation test. Grounded in social cognitive theory and control-value theory, this study aimed to estimate the overall correlation magnitude and test four potential moderators via subgroup analysis and Knapp-Hartung-corrected mixed-effects meta-regression: grade level, geographic region, academic self-efficacy measurement instruments, and test anxiety measurement instruments. Random effects models based on Fisher's Z transformation revealed a significant moderate negative correlation (r = -0.359, 95% CI [-0.405, -0.312]). Subgroup analyses and meta-regression indicated non-significant moderating effects for grade level, geographic region, and test anxiety measurement instruments, whereas academic self-efficacy measurement instruments significantly moderated the association. However, this moderating effect should be treated as preliminary suggestive evidence. Some subgroups comprised only one or two studies, so results should be interpreted with caution. Theoretically, this study is the first large-sample meta-analysis on this topic among Chinese secondary school students, providing synthesized empirical evidence for research on adolescent academic emotions and enriching the validation of social cognitive theory and control-value theory in the Chinese context. Practically, it offers tentative, hypothesis-generating guidance for selecting academic self-efficacy measurement instruments and proposes potential intervention directions for teachers, school counselors, and educational policymakers.