Skip to main content
← allmeta
Small-study-effect tests
Four small-study tests beyond Egger: Peters' regression for binary outcomes, Begg-Mazumdar rank correlation, Harbord's modified Egger, and Thompson-Sharp. Each has different power and bias properties; report multiple.
Input
Data format
Effect + SE (continuous/generic)
2×2 table: events1, n1, events2, n2
Studies (one per line)
Run tests
Results
Test Statistic p-value Applicable to Verdict
Funnel + regression overlay
Tests: Egger (continuous), Begg-Mazumdar rank correlation, Harbord-Egger modification for binary OR with log-score transform, Peters' test for log-OR using 1/(total sample size) as predictor, Thompson-Sharp with multiplicative dispersion. See Sterne JAC, Sutton AJ, Ioannidis JPA, et al. BMJ 2011;343:d4002 for funnel-plot recommendations and power comparisons (Egger 1997 is the original Egger paper, not a co-author of the BMJ 2011 article).