The ϒ – Ξ plane

ϒ = true scatter / quoted width — is the error bar honest? ·  Ξ = quoted width / exact-analysis width — was all the information used? Honest data cannot beat the exact analysis: ϒ·Ξ ≥ 1.

Drag the point — or a slider

Analysis dials

Sliders stop at the information bound ϒ·Ξ = 1. “Systematic part” turns scatter into a fixed offset of every estimate.

Paper anchors — click to load

Repeated experiments — quoted 68% bars vs the truth

What a PP plot would show