Game-based learning and achievement in undergraduate statistics and probability: the role of learning approach

This study examined undergraduate achievement in statistics and probability under game-based learning (GBL) and traditional instruction, while accounting for baseline achievement and learning approach. A quantitative quasi-experimental pre-test-post-test comparison-group design was used with 305 science-stream undergraduates enrolled in a Malaysian public university. Participants were recruited through convenience sampling and remained in five intact classes, which were randomly allocated at the class level using a random-number generator, with three classes assigned to GBL (n = 180) and two classes to traditional instruction (n = 125). Achievement was measured using a researcher-developed statistics and probability test, and learning approach was assessed using the R-SPQ-2F. Following a four-week intervention, a 2 × 2 factorial ANCOVA tested post-test achievement by pedagogical approach and learning approach, with pre-test achievement as the covariate. Pre-test achievement was significantly associated with post-test performance, F(1, 300) = 117.961, p < 0.001, η_p^2 = 0.282. After adjustment, pedagogical approach, F(1, 300) = 59.679, p < 0.001, η_p^2 = 0.166, and learning approach, F(1, 300) = 27.200, p < 0.001, η_p^2 = 0.083, were significant. The Pedagogical Approach × Learning Approach interaction was also significant, F(1, 300) = 17.210, p < 0.001, η_p^2 = 0.054. The GBL-traditional adjusted difference was larger in the Surface Approach group (16.50 points) than in the Deep Approach group (9.20 points). GBL was therefore associated with higher baseline-adjusted achievement. However, the small number of randomly allocated intact classes constrains causal interpretation, and the single-institution design limits generalisability.