Exploring the moderating role of group engagement on instructional methods and academic performance: a deep learning approach

Purpose This study aims to investigate the impact of traditional, blended, and collaborative instructional methods on academic performance in management education with a focus on the moderating role of group engagement. Design/methodology/approach The paper opted for an experimental study considering a sample of 54 Master of Business Administration students. The students were taught using traditional, blended, and collaborative methods over a semester. Group engagement was measured using a computer vision-based deep learning model. The deep learning model utilized facial expressions to estimate group engagement level, and academic performance was assessed through proctored exams. Grounded in the Stimulus-Organism-Response framework, the moderation analysis was performed to determine the effect of group engagement on the relationship between instructional methods and academic performance. Findings The results indicate that blended and collaborative learning significantly improve academic performance compared to traditional methods. High group engagement was found to positively moderate the effect of blended learning on academic performance, whereas its moderating effect on collaborative learning was not significant. Analysis of sample classes shows that students' emotions shift depending on the learning method, offering insights into engagement patterns. Originality/value This study innovatively applies a computer vision model, offering an objective assessment method for group engagement in educational settings. It also contributes to understanding how instructional methods and group engagement impact academic performance, with implications for instructional design in higher education.