<![CDATA[This study aims to describe students' research report writing skills before and after using ChatGPT as an academic reviewer, and to test the statistical and practical significance of its effect. A quantitative approach with a quasi-experimental one-group pretest-posttest design was employed. Thirty-six Grade XI.F3 students of SMAN 1 Barru were selected through purposive sampling. The instrument was a five-aspect rubric (structure & organization, content & argumentation, academic language, technical aspects, and originality & academic integrity; maximum score 60), validated through expert judgment and demonstrating excellent inter-rater reliability (ICC pretest = 0.930; ICC posttest = 0.910). At the posttest stage, students completed three sequential steps: independently rewriting a research report, submitting it to ChatGPT for academic-reviewer feedback, and then independently revising based on those recommendations. Data were analyzed using descriptive statistics, Shapiro-Wilk normality test, Paired Samples T-Test, and Cohen's d effect size. Mean scores improved from 79.0 (SD = 4.76) at pretest to 90.4 (SD = 3.0) at posttest. The Paired Samples T-Test yielded t (35) = −13, p = 0.000, rejecting H₀. Cohen's d = 2.76 indicated a very large effect. The greatest gains were in Content & Argumentation (+2.8 points) and Technical Aspects (+1.8 points), while the smallest gain was in Originality & Academic Integrity (+0.4 points). These findings confirm that structured ChatGPT integration with prompt-engineering training and academic-ethics reinforcement significantly enhances high school students' research report writing quality.]]>
Copyrights © 2026
