Implementation of the ARKA-STEM Model and Students’ Mathematical Critical Thinking: An Exploratory Two-Class Quasi-Experimental Study

Source abstract

This exploratory quasi-experimental study examined mathematical critical thinking in two intact Grade XII classes at SMAN 3 Sampit, Central Kalimantan: one taught with the ARKA (Activity, Reflection, Conceptualization, Application) model integrated with STEM (n = 33) and one taught with conventional direct instruction (n = 33), on the topic of Circle Equations and Their Elements. Because one class represented each condition, treatment is confounded with class membership; all inferential statistics are therefore reported as descriptive contrasts between these two classes and not as estimates of a population treatment effect. Critical thinking was measured with a six-item constructed-response test based on Facione’s framework (one item per indicator; 0–24 scale), scored by the researcher without blinding. In the ARKA-STEM class the mean score rose from 11.06 to 17.36 (gain = 6.30, 95% CI [5.62, 6.98]; mean N-Gain = 0.510), and in the comparison class from 9.82 to 11.27 (gain = 1.45; mean N-Gain = 0.112). Adjusted for pretest, the ARKA-STEM class scored 4.96 points higher (95% CI [4.08, 5.84]); because regression slopes differed between classes (p = .036), the conditional difference was larger at lower pretest scores. The standardized posttest difference was d = 1.59 (95% CI [1.04, 2.14]). Posttest reliability in the ARKA-STEM class was low (α = .531), so indicator-level results are exploratory. Observed implementation fidelity rose from 71.02% to 97.16% across three meetings. The ARKA-STEM class scored substantially higher than the comparison class under these conditions, but the design cannot establish that the model caused the difference; replication with several independently assigned classes, blinded scoring, and a validated instrument is required.