A three-level meta-analysis of the association between students' epistemic cognition and learning outcomes in South...

In the current era where unverified and non-credible information is prevalent, students' epistemic cognition has become increasingly critical in education. Epistemic cognition helps students develop the abilities to acquire, understand, justify, change, and use knowledge. Although the literature has provided extensive insights into epistemic cognition, the association between such cognition and learning outcomes remains unclear within specific cultural contexts. Given that epistemic cognition is particularly shaped by the culture, the purpose of this study was to synthesize empirical evidence on the association between students' epistemic cognition and learning outcomes in South Korea through a three-level meta-analysis.

Following the PRISMA guideline, this study conducted a three-level meta-analysis. A hierarchical model was employed to account for statistical dependence across 512 effect sizes from 18 peer-reviewed studies. Multivariate meta-regression and sensitivity analyses were performed to determine significant moderators and ensure the robustness of the pooled effect sizes.

The analysis yielded a statistically significant small-to-moderate pooled effect size (r = 0.191, p < 0.001). The heterogeneity test exhibited low-to-moderate variance at Level 2 (τ²  = 0.015, I²  = 35.82%) and moderate variance at Level 3 (τ²  = 0.024, I²  = 56.64%). The school level and epistemic cognition construct were found to be key moderators of the pooled effect size, while the other four, domain specificity, alignment, scale polarity, and type of learning outcome, were not.

These findings suggest that epistemic cognition is a critical factor in relation to different types of learning outcomes, such as higher and lower order cognitive outcomes, self-regulation, and motivation. The school level and epistemic cognition constructs appear to influence the association between epistemic cognition and learning outcomes. However, caution should be taken in interpreting these findings, given that the effect sizes were based on correlation coefficients from Korean studies, limiting causal inference and generalizability across different contexts and cultures.