The Engagement–Learning Paradox: A Mixed-Methods Study Toward Reflective Gamified Learning in Medical Education

Background: Although gamification can enhance learner motivation and engagement, increased engagement does not consistently translate into improved conceptual learning. This discrepancy may be particularly relevant for novice medical learners processing conceptually demanding biomedical content within limited instructional time. Objective: This study examined whether embedding a gamified activity within a structured instructional sequence was associated with higher post-intervention knowledge performance than gamification alone and explored how quantitative and qualitative findings could explain differences in learners’ experiences and outcomes. Context: The study was conducted among first-year medical students learning plasma membrane structure and transport at Universidad de La Sabana, Colombia. Methods: An explanatory sequential mixed-methods quasi-experimental design (QUAN → qual) was conducted with 239 first-year medical students allocated by cohort to a gamification-only (n = 115) or blended learning condition (n = 124). Knowledge acquisition was assessed via multiple-choice questions and learner experience via the GAMEFULQUEST instrument. Quantitative analyses included reliability indices, EFA, CFA, structural equation modeling (SEM), latent profile analysis (LPA), and psychometric network analysis (EBICglasso). Qualitative data were analyzed through Reflexive Thematic Analysis (Braun & Clarke) with independent dual coding and interpretative integration via joint displays. Results: Three weeks after the intervention, the blended-learning group showed a higher proportion of correct responses than the gamification-only group (88.7% vs. 65.2%; χ2 = 18.80, p < 0.001; OR = 4.19; NNT ≈ 5). SEM showed positive independent associations of Motivation (β = 0.451, p < 0.001) and Autonomy (β = 0.351, p < 0.001) with Perceived Learning, whereas Immersion (β = 0.051, p = 0.384) and Cognitive Load (β = −0.062, p = 0.166) showed no statistically significant independent associations. The structural model showed CFI = 0.982, TLI = 0.940, RMSEA = 0.112 (90% CI [0.051, 0.178]), and SRMR = 0.063; given the very low degrees of freedom (df = 3), RMSEA was interpreted cautiously. LPA identified heterogeneous learner profiles, while qualitative findings contextualized the quantitative patterns by highlighting reflection, self-regulation, and instructional scaffolding as salient features of students’ learning experiences. Conclusions: Gamification operates through motivational-regulatory architecture rather than immersion. Meaningful learning emerges when engagement is supported by autonomy, reflection, and instructional scaffolding, substantiating the Reflective Gamified Learning framework.