Funnel Plot Calculator

A funnel plot is a scatter plot used in meta-analyses to visually detect publication bias. Each study is plotted with its effect size on the x-axis and a measure of precision (typically the standard error) on the y-axis. In an unbiased meta-analysis, the points should form a symmetrical inverted funnel shape around the pooled effect estimate. Asymmetry suggests that smaller studies with non-significant results may be missing from the literature.

Most methodologists recommend at least 10 studies before interpreting a funnel plot, as smaller numbers of studies make it very difficult to distinguish true asymmetry from random variation. Egger's test in particular has low statistical power with fewer than 10 studies, so results should be interpreted with caution below that threshold.

Egger's test is a regression-based statistical test for funnel plot asymmetry. It regresses the standardized effect estimate against its precision (1/SE). A statistically significant intercept (p < 0.05) suggests asymmetry in the funnel plot, which may indicate publication bias or other small-study effects such as heterogeneity or methodological differences between large and small studies.

Conventionally, the effect size (e.g., log odds ratio, SMD) is placed on the x-axis and the standard error (or its inverse, precision) on the y-axis. Plotting SE with the scale inverted (larger at the bottom) creates the characteristic funnel shape where large, precise studies cluster near the top and small, imprecise studies spread out at the bottom.

An ideal funnel plot shows a symmetrical, inverted funnel shape. Studies with small standard errors (high precision) cluster tightly near the pooled effect at the top of the plot, while studies with larger standard errors spread symmetrically on both sides near the bottom. Perfect symmetry suggests no publication bias, though it does not rule out all forms of bias or confounding.

Asymmetry means the studies are not evenly distributed around the pooled effect estimate. This can be caused by publication bias (positive studies more likely to be published), small-study effects, between-study heterogeneity, chance (especially with few studies), or fraud. It does not automatically confirm publication bias — clinical and methodological context must also be considered.

Fixed-effect models assume all studies estimate the same true effect and differences are due to sampling error alone. Random-effects models (such as DerSimonian-Laird) assume the true effect varies between studies and incorporate between-study variance (tau²) into the weights. Random-effects models give relatively more weight to smaller studies, which can affect the shape and interpretation of the funnel plot.

No. A symmetrical funnel plot reduces the suspicion of publication bias but cannot definitively rule it out. Bias could be present but balanced, or the number of studies may be too small to detect asymmetry. Funnel plots should always be interpreted alongside other evidence such as trial registries, systematic search strategies, and Egger's or Begg's test results.