ChatGPT Use and Self-Reported Academic Writing Development among EFL Students: An Explanatory Sequential Mix-Methods Study

Abstract

This study examined the association between ChatGPT use and self-reported academic writing development and explored students’ perceptions of its role across the planning, drafting, revising, and editing stages. A quantitative-priority explanatory sequential mixed-methods design was employed at one university in Batam, Indonesia. Using purposive sampling, questionnaire data were collected from 80 English Education students who had completed an academic writing course and used ChatGPT for writing tasks. Eight respondents with different ChatGPT-use profiles subsequently participated in semi-structured interviews. The quantitative results showed that higher composite ChatGPT-use scores were significantly associated with higher self-reported academic writing scores, explaining 60.1% of the variance. Based on the eight interviews, participants perceived ChatGPT as a source of support for idea generation, text organization, language refinement, revision, efficiency, and confidence. They also emphasized the importance of verifying information, evaluating AI-generated suggestions critically, maintaining authorship, and avoiding excessive dependence. These findings indicate an association between ChatGPT use and perceived writing outcomes rather than objectively measured or causal improvement. The interpretation is limited by the one-institution purposive sample, cross-sectional design, and reliance on self-reported data. Pedagogically, the findings suggest that verification practices, critical evaluation, academic integrity, and learner autonomy warrant further consideration in AI-assisted writing instruction.