Meta Publication Bias: Funnel Plot + Egger's Test
Assess whether a meta-analysis has publication bias / small-study effects: draw the funnel plot (effect size vs standard error) and run Egger's linear-regression test of funnel symmetry. A clearly asymmetric funnel and an Egger intercept significantly different from 0 suggest possible publication bias. Computed locally in your browser; data are not uploaded. Use it alongside the meta-analysis tool.
① Enter the studies
One study per row: effect size standard error (space- or comma-separated). For ratio effects (OR/RR/HR), first take the natural log (lnOR, etc.) and use the SE on the log scale.
How to use & methodology
How do I read the funnel plot?
Ideally the study points scatter symmetrically around the pooled effect in an inverted-funnel shape (high-precision studies cluster at the top, low-precision ones spread at the bottom). If one side of the bottom is clearly missing or shifted, small negative studies may be unpublished (publication bias).
What does Egger's test measure?
It regresses each study's standardised effect (effect/SE) on its precision (1/SE) and tests whether the intercept is 0. A significantly non-zero intercept indicates funnel asymmetry — a quantitative measure that is fairly sensitive to publication bias.
How do I enter ratio effects?
OR, RR, HR and other ratios are symmetric only on the log scale, so first take the natural log (lnOR/lnRR/lnHR) and use the SE on the log scale (often back-calculated from the 95% CI: SE=(lnUL−lnLL)/3.92).
How many studies are needed?
Generally include ≥10 studies before running the funnel plot and Egger's test; otherwise the power is low and misjudgment is easy. With too few studies, just describe rather than force a conclusion. Asymmetry may also reflect true heterogeneity rather than bias.