Effects of teacher-guided ChatGPT-supported social studies instruction on middle school students' critical-thinking disposition: A quasi-experimental study

This quasi-experimental study examined whether teacher-guided, ChatGPT-supported social studies instruction was associated with middle school students' critical-thinking disposition. Participants were 56 seventh-grade students in two intact classes, with 28 students in the experimental class and 28 in the comparison class. The planned instructional sequence comprised six curriculum-aligned units delivered across six instructional weeks between March 24 and May 23, 2025, within a nine-week calendar window that included three weeks without a study unit because of the scheduled spring break, the week of the April 23 National Sovereignty and Children's Day holiday, and school midterm examinations. The experimental class used ChatGPT at predefined points in otherwise parallel lesson plans; the comparison class followed corresponding plans with the ChatGPT-specific components removed. Both classes were taught by the same social studies teacher. Critical-thinking disposition was assessed before and after the instructional sequence. Because the experimental class had a significantly higher pretest mean, posttest scores were compared using ANCOVA with pretest score as a covariate. The adjusted between-class difference was statistically significant, F(1, 53) = 8.013, p = .007, partial η 2 = 0.131, with an adjusted mean difference of 4.863 points, 95% CI [1.42, 8.31]. HC3 regression yielded b = 4.864, SE = 1.865, t(53) = 2.608, p = .012, 95% CI [1.124, 8.603]. Because each condition was represented by only one intact class, condition was completely confounded with class membership; the result therefore describes an adjusted difference between these two classes and does not identify an independent causal effect of ChatGPT or of the instructional package.