Funnel Plot
A scatter plot used to detect publication bias in meta-analyses — it plots effect size against precision, forming a symmetric funnel when bias is absent and an asymmetric one when it’s present.
// 01 — The chart
What it looks like
A funnel plot showing 12 studies. Most points cluster symmetrically around the pooled estimate (dashed vertical line). One low-precision study in the lower right suggests possible publication bias or an outlier.
// 02 — Definition
What is a funnel plot?
A funnel plot is a simple scatter plot designed to detect publication bias — the tendency for studies with positive or significant results to be published more often than those with null or negative findings. It’s one of the most important diagnostic tools in meta-analysis.
Each study is plotted as a point. The x-axis shows the effect size (odds ratio, risk ratio, mean difference), and the y-axis shows a measure of precision — typically the standard error (inverted so larger studies appear at the top) or sample size.
The key insight is geometric: large studies (high precision) cluster tightly at the top near the true effect, while small studies (low precision) scatter more widely at the bottom. This naturally creates an inverted funnel shape. If the funnel is symmetric, there’s no evidence of bias. If it’s asymmetric — with a gap where small negative studies should be — publication bias is suspected.
Origin: The funnel plot was introduced by Richard Light and David Pillemer in 1984 in their book Summing Up: The Science of Reviewing Research. It was further developed by Matthias Egger, George Davey Smith, and colleagues in the late 1990s, who proposed formal statistical tests for funnel plot asymmetry.
// 03 — Anatomy
Parts of a funnel plot
// 04 — Usage
When to use it — and when not to
- You need to assess publication bias in a meta-analysis
- Your meta-analysis includes at least 10 studies (fewer makes assessment unreliable)
- You want a visual complement to formal tests like Egger’s or Begg’s test
- Comparing study precision against effect magnitude across a body of evidence
- Identifying potential outlier studies that fall outside the expected funnel
- Presenting bias assessment alongside a forest plot in a systematic review
- You have fewer than 10 studies — too few to judge symmetry visually
- Your data isn’t from a meta-analysis (funnel plots are specific to this context)
- All studies are similarly sized — there’s no variation in precision to form a funnel
- You want to show results — use a forest plot instead; funnel plots assess bias, not effects
- The effect measure varies across studies and isn’t comparable on a single scale
- You rely solely on the funnel plot — always pair with formal statistical tests
// 05 — Reading guide
How to read a funnel plot
Follow these steps whenever you encounter a funnel plot in the wild.
Identify the axes
The x-axis shows the effect size and the y-axis shows precision (usually 1/SE or SE inverted). Larger, more precise studies appear near the top; smaller, less precise studies scatter near the bottom.
Find the vertical reference line
This dashed line represents the pooled effect estimate from the meta-analysis. In the absence of bias, studies should scatter symmetrically around it.
Assess the overall funnel shape
Look at whether the scatter of points forms a roughly symmetric, inverted funnel. The top should be narrow (precise studies clustering tightly) and the bottom should be wide (imprecise studies scattering broadly).
Check for asymmetry
If one side of the funnel has noticeably more points — especially if small studies with non-significant results are missing from one side — this suggests publication bias. The “gap” is the key signal.
Look for outliers outside the funnel
Points falling well outside the pseudo-confidence interval lines may represent studies with unusual methodologies, different populations, or data errors. These warrant closer inspection.
// 06 — Pitfalls
Common mistakes
Interpreting asymmetry as proof of publication bias
Fix: Funnel plot asymmetry can also be caused by genuine heterogeneity, methodological differences between small and large studies, or chance. Always pair visual inspection with formal statistical tests (Egger’s, Begg’s).
Using a funnel plot with too few studies
Fix: With fewer than 10 studies, there simply aren’t enough points to judge symmetry. The plot will appear sparse and any apparent asymmetry may be due to chance rather than bias.
Plotting sample size instead of standard error
Fix: While sample size can be used on the y-axis, standard error (or its inverse) is preferred because it directly reflects the precision of the effect estimate, producing a more interpretable funnel shape.
Ignoring the funnel boundaries
Fix: The pseudo-confidence interval lines (typically 95%) define the expected scatter. Without them, it’s hard to tell whether a point is an outlier or falls within normal sampling variation.
Not considering other causes of asymmetry
Fix: Small-study effects, true heterogeneity, and artefactual causes (poor methodological quality in small studies) can all produce asymmetric funnels. Publication bias is only one explanation.
// 07 — In the wild
Real-world examples
Cochrane systematic reviews
Every Cochrane review with 10 or more studies includes a funnel plot alongside the forest plot. Reviewers use it to flag potential publication bias and discuss whether missing studies might alter the conclusions of the meta-analysis.
Antidepressant efficacy meta-analyses
Funnel plots in psychiatric research have famously revealed publication bias in antidepressant trials. Studies submitted to the FDA showed a symmetric pattern, but published literature was asymmetric — negative trials were less likely to be published, inflating apparent efficacy.
Educational intervention reviews
The Campbell Collaboration uses funnel plots to check whether published evaluations of educational programs disproportionately report positive outcomes. Asymmetry has been found in tutoring and mentoring program evaluations.
// 08 — Quick reference
Key facts
// 09 — Variations
Types of funnel plots
The basic funnel plot has several enhancements that provide additional diagnostic power.
Standard funnel plot
The classic version with effect size on x and SE (or 1/SE) on y. Pseudo-CI lines form the expected funnel boundaries.
Contour-enhanced funnel plot
Adds shaded regions for statistical significance levels (p < 0.01, 0.05, 0.10). Helps distinguish publication bias from other causes of asymmetry.
Trim-and-fill funnel plot
Imputes hypothetical “missing” studies (shown as open circles) to make the funnel symmetric, then recalculates the adjusted pooled estimate.
Galbraith (radial) plot
An alternative that plots standardized effects against precision. Outliers are easier to identify, and the expected pattern is a horizontal band rather than a funnel.
// 10 — FAQs
Frequently asked questions
What is a funnel plot?+
A funnel plot is a simple scatter plot designed to detect publication bias — the tendency for studies with positive or significant results to be published more often than those with null or negative findings. It's one of the most important diagnostic tools in meta-analysis.
When should you use a funnel plot?+
Use a funnel plot when you need to assess publication bias in a meta-analysis. It also works well when your meta-analysis includes at least 10 studies (fewer makes assessment unreliable), and when you want a visual complement to formal tests like Egger’s or Begg’s test.
When should you avoid a funnel plot?+
Avoid a funnel plot when you have fewer than 10 studies — too few to judge symmetry visually. It is also a poor fit when your data isn’t from a meta-analysis (funnel plots are specific to this context), or when all studies are similarly sized — there’s no variation in precision to form a funnel.
Is a funnel plot suitable for dashboards?+
Yes — a funnel plot can work well in dashboards as long as the panel is large enough for readers to perceive the encoded values, has a clear title, and includes the legend or axis labels needed to interpret it.
What category of chart is a funnel plot?+
Funnel Plot belongs to the Statistical family of charts. Charts in that family are designed to answer the same kind of question, so they often work as alternatives when one doesn't quite fit your data.
How do you read a funnel plot?+
Start with the axis labels and legend, then look at the overall shape before zooming into individual marks. Compare prominent features against the rest of the data, and verify any conclusion against the underlying numbers when precision matters.