The Influence of Gamified Learning on Student Motivation: The Mediating Role of Engagement and the Moderating Effect of Self-Efficacy

Abstract

Gamified learning is widely reported to raise student motivation, and engagement is the mechanism most often credited with carrying that effect. This study tests whether engagement accounts for the relationship and whether the relationship holds equally for all learners. A conditional process model was specified in which engagement mediates the association between gamified learning and student motivation and academic self-efficacy moderates it. Four constructs were measured with ten items each, contextually adapted from four published instruments — the Gameful Experience Questionnaire, the student form of the Utrecht Work Engagement Scale, and the self-efficacy and motivation scales of the Motivated Strategies for Learning Questionnaire — and content-validated before administration by a panel of seven experts, which returned S-CVI/Ave =.903 and retained all forty items without revision. An online questionnaire returned 356 complete responses from university students in technology-supported courses, and the model was estimated by partial least squares structural equation modelling with 5,000 bootstrap resamples. All three hypotheses were supported. Gamified learning was positively associated with motivation (β =.277, p <.001), engagement mediated part of that association (β =.114, 95% CI [.057,.175]), and self-efficacy moderated the direct path (β =.171, p <.001) while also predicting motivation directly (β =.194, p <.001). Because the direct and indirect effects were both significant and shared a sign, the pattern is complementary partial mediation, in which the mediated route carries 29.1% of a total effect of.391. The contribution is to bound the engagement account rather than to confirm it: the direct effect surviving the mediator is more than twice the size of the mediated effect, so most of what gamified design does to motivation is not explained by raising engagement as measured here. The design is cross-sectional, and the sample was recruited by voluntary response, so the results are reported as associations within a conditional process model rather than as causal effects.