Explore how publication bias distorts scientific findings.
Balls fall through rows of pegs. At each peg, a ball randomly goes left or right — producing a bell-curve distribution across the bins, just like the spread of results across many experiments where the true effect is zero.
When Selection on Significance is on, only results in the highlighted red region are “published” — modeling publication bias. The filtered mean drifts away from 0, showing how the scientific literature can systematically overestimate effect sizes when non-significant results go unreported.