Relative Effectiveness of Gamified Instructional Delivery Modes and Design Components on Student Academic Achievement: A Bayesian Network and Component Network Meta-Analysis

Empirical evidence on the impact of educational gamification on academic achievement varies substantially, yet the roles of instructional delivery modes (IDMs) and design components remain underexplored. Following PRISMA-NMA standards, this Bayesian network meta-analysis (NMA) established an evidence-based hierarchy of IDMs. Of 147 eligible studies, one extreme outlier was excluded, leaving 146 studies (2013–2025; N = 14,979; 177 comparisons) comparing gamified and non-gamified conditions across face-to-face (F2F), blended (BLN), and online (ONL) IDMs. Gamified blended instruction (GBLN) emerged as the most effective mode (g = 1.06, 95% CrI [0.529, 1.593]; SUCRA = 0.933), followed by gamified F2F instruction (GF2F; g = 0.74, 95% CrI [0.436, 1.045]; SUCRA = 0.712). Non-gamified modes showed no credible differences, indicating that delivery affordances were operationalized via gamification. Component NMA showed stable positive estimates for sensation across modes. Mode-specific estimates were credibly positive for cooperation in GF2F and GBLN and for time pressure in GF2F and GONL, although the corresponding between-mode contrasts did not exclude zero. Storytelling tended to decrease achievement across modes. Chance was credibly negative in GF2F, and its GONL–GF2F contrast was the only between-mode difference to exclude zero, an exploratory finding given data sparsity. Component × delivery mode interaction modeling explained 10.2% of the heterogeneity, whereas domain-level taxonomies explained near-zero variance because opposing elements masked one another. Despite a “very low” CINeMA certainty rating owing to high heterogeneity, sensitivity analyses confirmed the top rank of GBLN. Gamification therefore requires strategic alignment between design components and delivery mode affordances rather than a one-size-fits-all approach.