Publication Bias in 3D
A 3D field of study-result spheres: most fall gray into the file drawer, rare glowing ones rise onto the published-literature shelf. Watch the file-drawer effect inflate the visible anomaly rate above the true rate as significance threshold, true anomaly rate and study count change.
This scene renders a growing field of instanced spheres, each one a completed scientific study. Most are noise and, correctly, find nothing — but a fixed share of noise studies still cross the significance threshold by chance, and those imposters join the rare genuine anomalies rising to a glowing "published literature" shelf, while every honestly-null result drops out of sight into the file drawer below. Rotate and orbit the scene to see how the shelf fills up mostly with false positives as the study count grows, exactly the file-drawer and multiple-comparisons dynamic documented in real meta-science.
Studies fall onto a lit 'published' shelf or into a dim 'file drawer', illustrating how many significant results are published while others remain unseen.
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install