Contents
Academy · Research Data, Analytics & AI · Ideal Funnel Chart Properties
The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot
A well-formed funnel chart is a symmetrical, inverted-funnel scatter plot of study precision against effect size, constructed from at least approximately ten studies with standard error on the vertical axis, rather than a decorative figure added retrospectively.
A funnel chart plots each included study's treatment-effect estimate against a measure of that study's precision, typically its standard error, to provide reviewers with a visual check on whether the body of evidence contributing to a meta-analysis appears complete. In the absence of bias and genuine heterogeneity, the scatter should take the shape of an inverted funnel: small, imprecise studies spread widely near the bottom of the plot, and larger, more precise studies clustering tightly near the top, symmetrically around the pooled effect. A well-formed funnel chart is defined not by preference but by a specific, checkable set of properties: correct axis choice, a large enough number of studies, a consistently defined effect measure, and often a set of reference lines or shaded regions, which together determine whether the plot can carry any diagnostic weight at all.
Learning objectives
- The visual and statistical properties that distinguish a usable funnel chart from a decorative one
- Why standard error, rather than raw sample size, is the recommended measure of precision for the vertical axis
- The minimum number of studies required before a funnel chart is worth constructing
Common misconceptions
- “Any scatter plot of effect size against sample size constitutes a funnel plot.” The foundational 1997 BMJ paper by Egger and colleagues and the 2001 follow-up by Egger and Sterne both specify that the value of the plot depends on using a measure of precision, the standard error, rather than sample size on the vertical axis, because sample size alone produces an unpredictable shape even when no bias is present.
Characteristics of a well-formed funnel chart
A well-formed funnel chart is, first and foremost, symmetrical around the pooled effect estimate once precision is plotted correctly. Egger and colleagues' 1997 paper, which introduced the now-standard graphical test for this kind of asymmetry, describes the ideal shape: in a body of evidence free of bias, the scatter of individual study results should resemble an inverted funnel, wide and dispersed near the bottom of the plot where imprecise, usually smaller, studies sit, and narrow and tightly clustered near the top where precise, usually larger, studies sit1. This shape follows directly from basic sampling theory: a small study's effect estimate is subject to more sampling error, so its point estimate can lie further from the true effect purely by chance, whereas a large study's estimate is constrained more tightly around the true effect. A funnel chart that instead shows a gap, a noticeably empty region on one side of the plot among the smaller, less precise studies, indicates that a possible cause, such as publication bias, selective reporting or a real difference in effect by study size, requires investigation.
Egger and Sterne's 2001 follow-up paper in the Journal of Clinical Epidemiology makes the axis choice explicit as part of this property: a funnel chart should use standard error, not sample size, as its precision measure, because plots built on sample size can take an unpredictable shape even when the underlying studies are entirely free of bias, which defeats the purpose of checking for a recognisable, interpretable funnel shape in the first place2. Choosing standard error keeps the expected shape consistent across different effect measures and different combinations of study designs, and it is this property that makes departures from it meaningful. A chart that uses the wrong axis can look asymmetric, or deceptively symmetric, for reasons that have nothing to do with the evidence base it is meant to represent.
Preconditions for a trustworthy funnel chart
A funnel chart only carries diagnostic weight once a meta-analysis includes a reasonable number of studies spanning a genuine range of precision; below that, the plot's shape is determined more by chance than by any genuine feature of the evidence. Sterne and colleagues' 2011 BMJ methods paper, now the standard reference for using and interpreting these plots, recommends against relying on any formal statistical test of funnel-plot asymmetry, such as Egger's test, when a meta-analysis has fewer than about ten included studies, because the test's statistical power to distinguish genuine asymmetry from ordinary chance variation is too low below that threshold3. Sterne, Gavaghan, and Egger's earlier empirical study of publication bias had already demonstrated the scale of this problem of statistical power: formal tests for small-study effects routinely lacked adequate power in the meta-analyses they examined, which is why visual inspection of the chart's shape, imperfect as it is, remains a necessary companion to any formal test rather than a redundant extra step7.
A second condition, less often stated explicitly but equally important, is that the included studies require a genuine spread of precision for the plot to be informative at all. Begg and Mazumdar's 1994 paper on the operating characteristics of their rank correlation test for publication bias found that both their test and Egger's regression-based test perform poorly when the included studies cluster tightly around a similar sample size, because a funnel shape, narrow or wide, cannot meaningfully form without a real range of precision to plot against8. Reviewers should therefore check, before building or interpreting a funnel chart, both that enough studies are available and that those studies differ sufficiently in size to produce an interpretable shape.
Design choices that enhance the usefulness of a funnel chart
Beyond the basic axis and study-count requirements, several specific design choices distinguish a funnel chart that supports a genuine judgement from one that merely appears authoritative. Peters and colleagues' 2006 JAMA paper introduced the contour-enhanced funnel plot, which overlays shaded regions corresponding to conventional statistical-significance thresholds (p<0.01, p<0.05, p<0.1) directly onto the chart4. This single addition allows a reviewer to see directly whether any gap in the funnel's expected shape falls specifically in the statistically non-significant region, the pattern publication bias is expected to produce, or elsewhere, which points toward a different explanation such as genuine heterogeneity. A chart without these contours requires the reader to infer significance mentally from the position of each point, which is both slower and more error-prone.
A second useful design choice is to mark the pooled effect estimate with a clear vertical reference line and, where a trim-and-fill style adjustment is reported alongside the chart, to distinguish the originally observed studies from Duval and Tweedie's imputed, hypothetically missing ones using a different marker shape or shade5. Lau and colleagues' 2006 BMJ editorial on the misuse of funnel plots cautions specifically against publishing a funnel chart that implies a firm conclusion through its visual design, such as bold highlighting of asymmetry or a single confident caption, when the underlying evidence base is too small or too heterogeneous to support one6. The chart's design should, therefore, reflect and communicate the actual strength of the underlying evidence, not overstate it. The Cochrane Handbook's own guidance on reporting biases situates the funnel chart within this larger judgement, instructing review authors to report the chart alongside an explicit statement of which alternative explanations for any observed shape were considered, rather than presenting the image as self-sufficient evidence of bias on its own9, a principle PRISMA 2020's reporting items echo by requiring any assessment of risk of bias from missing results to be described in the methods and results separately from the figure itself10.
Common misinterpretations that a well-constructed chart cannot prevent
Even a chart that satisfies every design condition above remains vulnerable to a specific misreading: treating any asymmetry it shows as proof of publication bias specifically, rather than as one of several possible causes. Sterne and colleagues' 2011 BMJ paper is explicit that genuine heterogeneity (smaller trials conducted in different populations or under different protocols from larger ones) and methodological weaknesses correlated with study size can both produce an asymmetric shape that has nothing to do with missing studies3. A well-formed chart narrows the plausible explanations and makes any gap easier to see; it does not, by itself, indicate which explanation applies to a specific gap. Lau and colleagues' 2006 editorial found that a majority of the BMJ reviews they examined omitted this step entirely, drawing a categorical conclusion from the chart's shape without first ruling out these alternatives6.
A second common misreading is the assumption that a chart built correctly at the level of axis choice and study count is therefore also built on a consistent effect measure throughout. Egger and Sterne's 2001 guidance notes that mixing effect measures, for example plotting some studies as odds ratios and others as risk ratios without converting them to a common scale first, can itself introduce an artificial asymmetry unrelated to any real property of the evidence2. Reviewers checking a published funnel chart for trustworthiness should therefore confirm not only the axis and study count, which are usually stated or inferable, but also that a single, clearly defined effect measure was used for every study plotted.
Key points
Key points
A funnel chart is informative in a meta-analysis only when it uses standard error on the vertical axis, draws on at least approximately ten studies with a genuine spread of precision, and is accompanied by an explicit account of the alternative explanations for its shape that were considered, rather than being treated as self-explanatory.
Practical application
Practical application
- The number of studies and the spread of precision should be checked before the chart is constructed. Approximately ten or more included studies with genuinely different sample sizes should be available before the shape of the chart is expected to carry diagnostic weight.
- Standard error, rather than sample size, should be plotted on the vertical axis. This keeps the expected funnel shape consistent and interpretable across different effect measures and study designs.
- Contour lines or shading for statistical significance thresholds should be added where the software permits. These allow direct assessment of whether any gap in the funnel lies in the non-significant region associated with publication bias.
Frequently asked questions
No. Symmetry is reassuring but not conclusive. A funnel chart lacks the power to detect bias reliably when a meta-analysis has few studies, and some patterns of selective reporting do not produce a visually detectable gap even when real.
Plotting raw sample size instead of standard error on the vertical axis is the most frequently cited design error, since Egger and Sterne's 2001 analysis showed sample size alone produces an unpredictable shape that is not interpretable in the same way.
Sterne and colleagues' 2011 guidance advises against relying on formal statistical asymmetry tests below roughly ten studies, though a chart can still be shown for context as long as readers are told explicitly that it should not be used to draw firm conclusions at that size.
Only when enough studies with a genuine spread of precision are available. A chart built from too few or too similar studies risks implying a diagnostic conclusion that the data cannot support.
References
- Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315(7109):629-634. read 10 October 2026. View on The BMJ
- Egger M, Sterne JAC. Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis. Journal of Clinical Epidemiology. 2001;54(10):1046-1055. read 10 October 2026. View on Journal of Clinical Epidemiology
- Sterne JAC, Sutton AJ, Ioannidis JPA, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ. 2011;342:d4002. read 10 October 2026. View on The BMJ
- Peters JL, Sutton AJ, Jones DR, Abrams KR, Rushton L. Comparison of two methods to detect publication bias in meta-analysis. JAMA. 2006;295(6):676-680. read 10 October 2026. View on JAMA
- Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000;56(2):455-463. read 10 October 2026. View on Biometrics
- Lau J, Ioannidis JPA, Terrin N, Schmid CH, Olkin I. The case of the misleading funnel plot. BMJ. 2006;333(7568):597-600. read 10 October 2026. View on The BMJ
- Sterne JAC, Gavaghan D, Egger M. Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature. Journal of Clinical Epidemiology. 2000;53(11):1119-1129. read 10 October 2026. View on Journal of Clinical Epidemiology
- Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994;50(4):1088-1101. read 10 October 2026. View on PubMed
- Sterne JAC, Egger M, Moher D (editors). Chapter 10: Addressing reporting biases. In: Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions, Version 5.1.0. The Cochrane Collaboration; 2011. read 10 October 2026. View on Cochrane Training
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. read 10 October 2026. View on The BMJ
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot. Pubrica. https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/
Pubrica Academy. “The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot.” Pubrica, 10 Oct. 2026, pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/.
Pubrica Academy (2026) The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot. Pubrica. Available at: https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/ (Accessed: 15 October 2026).
Pubrica Academy. “The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot.” Pubrica. October 10, 2026. https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/
Pubrica Academy, ‘The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/> accessed 15 October 2026.
Pubrica Academy. The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/
Pubrica Academy, “The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/
@misc{/la_ideal_funnel_chart_meta_analysis_2026,
author = {{Pubrica Academy}},
title = {The Expected Appearance of a Funnel Chart in Meta-Analysis: Criteria for a Well-Formed Plot},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/ideal-funnel-chart-in-a-meta-analysis/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Funnel Chart Build Checklist
This checklist should be completed before a funnel chart for a meta-analysis is constructed or published.
[ ] Confirmed at least ~10 studies with a genuine spread of precision are included [ ] Plotted standard error, not raw sample size, on the vertical axis [ ] Marked the pooled effect estimate with a clear reference line [ ] Added contour/shading for significance thresholds if software allows [ ] Distinguished observed studies from any trim-and-fill imputed studies [ ] Documented which alternative explanations for the chart's shape were considered
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica’s meta-analysis team constructs funnel charts to current methodological standards, with the correct axis, an adequate number of studies and contour enhancement where useful, and accompanies every chart with a written account of what its shape does and does not support.