A funnel plot is a scatter of each study’s effect estimate against its standard error, used to screen a meta-analysis for small-study effects and possible publication bias. This funnel plot generator centres the funnel on the pooled estimate, draws the diagonal pseudo 95% confidence limits, and plots your studies so you can judge whether the scatter is symmetric, then download the figure for your manuscript.
Funnel plot generator
Paste your studies and get a funnel plot centred on the pooled estimate, with the pseudo 95% confidence limits drawn, ready to download as an SVG.
Drag and drop or click. CSV, TSV, Excel; header row with columns like study, label, author, trial, estimate.
Asymmetry tests
Egger's regression intercept0.19 (p = 0.9278)
Begg's rank correlation (Kendall's tau)0.14 (p = 0.6523)
Egger's test is not below the 0.05 threshold, so it does not suggest funnel plot asymmetry.
Begg's test is not below the 0.05 threshold, so it does not suggest funnel plot asymmetry.
Both tests have low power with fewer than ten studies, so read them alongside the plot itself.
7 studies plotted against the fixed-effect pooled estimate and pseudo 95% limits.
Report-ready text
Visual inspection of the funnel plot together with Egger's test (intercept = 0.19, p = 0.9278) and Begg's test (Kendall's tau = 0.14, p = 0.6523) did not indicate small-study effects.
What the funnel plot is actually testing
The logic of a funnel plot is precision. A large, precise study should land close to the true effect, while a small, imprecise one can fall some distance either side of it by chance alone. Plot effect against standard error and that expectation has a shape: a wide base of scattered small studies narrowing to a tight neck of precise ones at the top. When studies are missing from one lower corner, the funnel looks lopsided, and the usual suspicion is that small studies with unwelcome results were never published. Our guide to reading a funnel plot for asymmetry works through the patterns in detail.
Asymmetry is a clue, not a verdict
A skewed funnel is consistent with publication bias in a meta-analysis, but it is not proof of it. Genuine clinical heterogeneity, where smaller trials were run in sicker or different populations, can bend the funnel just as effectively. That is why the plot is a screening step, paired with a formal test rather than read alone. Quantify the asymmetry with Egger’s regression test before you write a word about bias, and treat the visual as the prompt to look closer.
A worked example and the traps to avoid
Imagine seven trials of a treatment, six clustered neatly around a risk ratio of about 1.3 and one small, imprecise trial sitting alone at 1.9 near the bottom of the plot. The scatter leans right, and the missing mirror-image study at the bottom left is what draws the eye. With only seven studies, though, that gap could easily be chance, and the honest reading is that the funnel raises a question rather than answers one. The plot here centres on the pooled estimate and draws the pseudo 95% limits so you can see at a glance how many points fall outside them.
Common mistakes researchers make
The first mistake is drawing a funnel plot with too few studies; below about ten points, asymmetry and chance are indistinguishable and the figure misleads more than it informs. The second is plotting effect against sample size instead of standard error, which distorts the funnel because precision, not headcount, is what should govern the spread. The third is jumping from a lopsided funnel straight to a claim of publication bias without considering heterogeneity, a leap that the wider meta-analysis workflow is designed to guard against. When the bias assessment has to satisfy reviewers, our statistical analysis service runs the plot, the tests, and the interpretation together.
How it works
Ratio measures are placed on the log scale; the axis labels are exponentiated back. The vertical axis is the standard error, drawn with the smallest value at the top. The funnel is centred on the inverse-variance fixed-effect pooled estimate.
pooled estimate = sum(yi / si^2) / sum(1 / si^2)
pseudo 95% limit at se s: estimate +/- 1.96 * s
The two dashed diagonals join the apex at standard error zero to those limits at the largest standard error, so a study outside the funnel is one whose estimate sits more than about 1.96 standard errors from the pool. Asymmetry should be confirmed with a formal test such as Egger’s before it is interpreted.
Frequently asked questions
- How do you make a funnel plot step by step?
- Gather every study's effect estimate with a confidence interval or standard error on one scale, then decide whether you are plotting a ratio or a difference measure. Paste the studies one per line; the generator pools them, draws the pooled estimate as the vertical spine, adds the two diagonal pseudo 95% confidence limits, and plots each study as a point at its effect against its standard error. Read the scatter for symmetry, then download the figure for your manuscript.
- How should a funnel plot look?
- The most precise studies, those with the smallest standard error, sit near the top close to the pooled estimate, and less precise studies scatter wider towards the bottom, so the cloud of points should resemble a symmetric inverted funnel. Roughly 95 percent of the points are expected to fall inside the diagonal confidence limits if there is no asymmetry. A balanced, symmetric spread is reassuring; a lopsided one is the cue to investigate.
- Can you make a funnel plot in Excel?
- It is possible with a scatter chart by plotting effect on one axis and standard error on a reversed axis, but you have to add the pooled spine and the diagonal confidence limits as separate manual series, and any change to the data breaks the geometry. This generator computes the pooled estimate and the pseudo 95% limits for you and exports a clean SVG that stays sharp at any size in a journal figure.
- How do you create a funnel plot for a meta-analysis?
- Use the same effect estimates and standard errors you pooled in the meta-analysis. The convention is to centre the funnel on the pooled estimate and put standard error on the vertical axis with the smallest values at the top, because precision, not sample size alone, governs how tightly a study should sit to the summary. Enter those values here and the plot is built to that convention automatically.
- What sample size is needed for a funnel plot?
- It is the number of studies, not the participants, that matters: a funnel plot is hard to read with fewer than about ten studies because asymmetry cannot be distinguished from chance, and most guidance discourages formal asymmetry tests below that count. Larger per-study samples help only in that they shrink each study's standard error and lift its point towards the top of the funnel, which sharpens the shape.
The guides behind this tool
More review and meta-analysis tools
Meta-analysisPool study effect estimates with fixed-effect and random-effects models, with the heterogeneity statistics shown.Open tool Heterogeneity (I-squared)Quantify how much your studies disagree: Cochran's Q, I-squared, H, and the between-study variance tau-squared.Open tool Odds ratio / risk ratioTurn a 2x2 table of events into an odds ratio, risk ratio, and risk difference, each with a 95% confidence interval.Open tool