<![CDATA[This study investigated the effect of AI-integrated STEM activities using cocoa leaf disease classification materials on undergraduate students’ algorithmic thinking skills. This quasi-experimental study used a non-randomized pretest–posttest comparison design and involved 66 undergraduate students. Data were collected using an algorithmic thinking test and an instrument validation sheet. The instrument demonstrated acceptable content validity, with a mean validation score of 4.57, a feasibility percentage of 91.38%, Aiken’s V coefficient of 0.892, and very high internal consistency (Experimental α = 0.942 and Control α = 0.938 ). The data were analyzed using assumption testing, Welch’s t-tests, gain-score analysis, and ANCOVA. The experimental group achieved a substantially larger mean gain than the control group (d = 1.43) and a higher average normalized gain (〈g〉 = 0.87 versus 0.52), and outperformed the control group on all seven algorithmic thinking indicators (d = 0.72 to 1.60). The ANCOVA indicated a significant group difference in posttest performance, F(1, 63) = 29.302, p < .001, with a higher adjusted mean in the experimental group (M = 76.8) than in the control group (M = 66.5). However, the pretest covariate explained virtually none of the posttest variance, so the covariance adjustment was negligible, and the two intact classes differed significantly at baseline. Because posttest variance was also heterogeneous, interpret the findings with appropriate caution and understand them as associational rather than causal. These results suggest that AI-integrated STEM activities based on cocoa leaf disease classification may support the development of undergraduate students’ algorithmic thinking through structured, iterative, and evidence-based learning processes. Keywords: algorithmic thinking, STEM activities, cacao diseases, classification, undergraduate students.]]>
Copyrights © 2026
