Back to top

Methodology

  • Introduction
  • Methodology
  • Interactive Visualization
  • Metric Results
  • Conclusion

    • Methodology

    The project follows a systematic review and meta‑analysis framework. For each study and outcome, we calculated standardized mean differences or mean differences with 95% confidence intervals, comparing mask‑on versus mask‑off conditions. We then applied random‑effects meta‑analyses to account for genuine differences between study populations, exercise tests and mask conditions.

    We quantified heterogeneity using the I² statistic and tau², and ran subgroup analyses by fitness level (healthy vs trained) and by mask type (surgical, FFP2/N95, cloth). To diagnose robustness and potential bias, we generated a suite of plots: forest plots for pooled effects, funnel plots for publication bias, L’Abbé‑style plots for direct mean comparisons, Baujat and radial plots for influential studies and heterogeneity, and bubble plots for simple meta‑regression.

    These visual tools allowed us not only to summarise overall effect sizes, but also to see where specific protocols, mask types or populations drove variation in the evidence.