← all 64 domains
domain 18 / 64 · MOMUS · the peer, the god of blame

MOMUS

the exterior witness; what each work owes its critic.
21 reviewers on the panel · the desk marks, then stamps · sealed & live
✎ THE REVIEW DESK · the panel marks the manuscript
UNDER REVIEW · —
hover a reviewer · click to enter its sphere

✎ the panel

every reviewer, named & enterable
MOM · THE PEERMOMUS · the exterior witness · 15 reviewsTHE PEER REVIEWmomus · three reviewers, three opinions — measure agreement THE REPLICATIONmomus · a published p<0.05 effect shrinks on replication THE NULL HYPOTHESISmomus · significant ≠ true — the false-discovery rate dependTHE GOODHARTmomus · when a measure becomes a target it ceases to be a goTHE STRESS TESTmomus · load to failure — the margin matters more than the rTHE HINDSIGHTmomus · outcome bias — the critic who judges by the result hTHE STEELMANmomus · attack the strongest form, not a strawman · lit+ambeTHE BLAMEmomus · Momus found fault in perfect works — fair critique vTHE AUDITmomus · you can't check everything — sample, and reportFLEISS' κmomus · agreement for a WHOLE panel of raters, corrected forKRIPPENDORFF'S αmomus · the most general reliability number — any raters, anTHE ICCmomus · for NUMBER ratings: share of variance that is real bTHE BLAND-ALTMANmomus · do two measurement methods AGREE? plot difference vsCRONBACH'S αmomus · do a scale's items hang together? internal consKENDALL'S Wmomus · do several judges RANK a set alike? concordance in [SPEARMAN'S ρmomus · do two reviewers' rankings move together? monotCOHEN’S κmomus · agreement beyond chance — raw agreement lies, κ subtTHE META-ANALYSISmomus · many studies pooled by inverse variance — a precise THE FUNNELmomus · publication bias — the missing side of the funnel inTHE MULTIPLE COMPARISONSmomus · test enough true nulls and you discover noise — FWER