<![CDATA[This study investigates whether essay-writing skills mediate the relationship between ChatGPT usage and critical thinking among undergraduate English as a foreign language (EFL) students. Using a quantitative design with partial least squares structural equation modeling (PLS-SEM) in SmartPLS, we analyzed data from 201 students at UINSA Surabaya (167 women; 34 men). PLS-SEM was selected for its suitability in modeling complex mediation with small to medium samples. The measurement model captured essay-writing elements (e.g., clarity, argumentation, coherence) and critical-thinking facets (e.g., inference, reasoning, evaluation). Structural results indicate that ChatGPT usage has a significant, positive association with critical thinking (path coefficient β = 0.401, p < .001). Essay-writing skills also exhibit a positive effect on critical thinking and function as a meaningful mediator, suggesting that gains in writing quality help translate the use of ChatGPT into higher-order thinking. Diagnostic checks flagged several cross-loadings, indicating potential threats to discriminant validity and construct clarity. Because unaddressed cross-loadings can bias parameter estimates and reduce measurement equivalence, educators and researchers should interpret structural paths with caution and align decisions with theory. Pedagogically, the findings support the guided integration of ChatGPT, emphasizing the critical appraisal of AI output, source evaluation, and iterative drafting, to strengthen students’ argumentation and reasoning. We recommend that instructors scaffold reflective prompts and verification routines so that perceived accuracy aligns with actual accuracy, thereby leveraging AI while safeguarding academic rigor.]]>
Copyrights © 2025
