Free scoping read. A written assessment of your manuscript in two business days, at no cost.

Start one
Contents

Academy · Research Data, Analytics & AI · Standard Error as the Y-Axis

Why Is Standard Error Preferred Over Total Sample Size on the Vertical Axis of a Funnel Plot?

Egger and Sterne’s 2001 methodological analysis found that plotting sample size rather than standard error on the vertical axis of a funnel plot produces an unpredictable shape even when no bias is present, which is why standard error became the recommended default. This article explains the mathematical basis for that recommendation and its practical consequences.

Pubrica AcademyWritten by a biostatistics writer Reviewed bya meta-analysis methodologist Published10 October 2026 Last checked10 October 2026 Reading time7 minutes Sources10 peer-reviewed

The vertical axis of a funnel plot is intended to represent the precision of each included study, and the choice of precision measure is not a cosmetic detail. Egger and Sterne’s 2001 paper in the Journal of Clinical Epidemiology, written specifically to resolve this question, concluded that the standard error should generally be used as the measure of study size on the vertical axis of a funnel plot, in preference to the total number of participants enrolled in a study2. The rationale for that recommendation is mathematical rather than stylistic: the standard error has a predictable statistical relationship to the effect estimate of a study that sample size does not, and it is this relationship that gives the funnel shape its diagnostic meaning.

Learning objectives

  • The mathematical reason that standard error produces a predictable funnel shape whereas sample size does not
  • How the interpretability of a funnel plot is compromised when sample size is used instead
  • The practical plotting advantages of the standard-error axis, namely confidence-interval lines and consistent scaling

Common misconceptions

  • “Sample size is the simplest and most intuitive representation of study precision on a funnel plot and is therefore an acceptable substitute for the standard error.” Egger and Sterne’s analysis found that a funnel plot based on sample size has an unpredictable expected shape even in a body of evidence entirely free of bias, which means that asymmetry on such a plot cannot be interpreted in the same way as asymmetry on a standard-error plot.

The mathematical basis for the standard-error axis

The standard error of a study is mathematically related to its effect estimate in a specific, well-understood way: for most common effect measures, the standard error decreases in a predictable, approximately inverse-square-root relationship as the number of participants increases, and this relationship holds consistently across different kinds of effect measure, such as mean differences, risk ratios and log odds ratios2. Because of this consistent relationship, a plot of standard error against effect size will, in a body of evidence with no bias and no true heterogeneity, form the same recognisable, symmetrical, inverted-funnel shape regardless of the specific effect measure used in the meta-analysis. This consistency is precisely what allows a reviewer to compare the shape of a given funnel plot with the expected general shape and to identify a departure from it as warranting investigation.

Sample size lacks this direct mathematical relationship to the effect estimate for a structural reason: the precision of a study depends not only on the number of participants enrolled but also on how those participants were distributed across comparison groups, the degree of outcome variability within the sample and the specific effect measure calculated from the data. Two studies with identical sample sizes may have meaningfully different standard errors if one had a more even allocation between treatment and control groups, or less outcome variability, than the other. Egger and Sterne concluded that this discrepancy between sample size and actual precision renders the expected shape of a sample-size-based funnel plot unpredictable, even under the ideal, bias-free scenario, which removes the basis for treating any particular shape as diagnostically meaningful2.

Consequences of plotting sample size on the vertical axis

The practical consequence of this mathematical discrepancy is that a sample-size-based funnel plot may appear symmetrical when genuine small-study effects are present, or asymmetrical when the evidence base is sound, for reasons unrelated to publication bias or heterogeneity. The 2011 BMJ methods paper by Sterne and colleagues, which consolidates the guidance of the field on these plots, treats the choice of precision measure as foundational to the rest of the analysis, noting that the validity of any subsequent asymmetry test depends on the vertical axis having been specified correctly3. An asymmetry test conducted on a plot with an inappropriate axis is not merely a weaker version of the correct analysis; it tests a different and less meaningful quantity.

The 2006 BMJ editorial by Lau and colleagues on the misuse of funnel plots found that inconsistent axis choices formed part of a broader pattern of technical problems undermining the use of these plots in published systematic reviews, alongside outcome coding and effect-metric choices that varied between reviews6. Because the choice of vertical axis is made early in the construction of the plot and is rarely scrutinised by readers thereafter, an incorrect choice may pass unnoticed through peer review into a published meta-analysis, which is one reason methodological guidance has moved towards specifying the standard error rather than leaving the choice to reviewer preference.

Practical plotting advantages of the standard-error axis

Beyond the underlying mathematics, the standard-error axis offers a concrete plotting advantage that a sample-size axis cannot replicate: it allows straight diagonal lines marking a 95% pseudo-confidence interval around the pooled effect to be drawn directly onto the chart, since those lines are themselves defined in terms of the standard error2. These reference lines give the reader an immediate visual indication of which plotted studies fall within the range of estimates expected by chance around the pooled effect and which fall notably outside it, and several funnel plot software packages include them by default once the standard-error axis is selected. A sample-size axis does not support such reference lines in any statistically meaningful way, because sample size does not correspond to the same confidence-interval mathematics.

The standard-error axis also weights the emphasis of the plot appropriately towards the studies most informative for detecting small-study effects. Because the axis compresses the visual space occupied by very large, highly precise studies and expands the space occupied by smaller, less precise studies, it directs the reader’s attention towards the region of the plot, that of the smaller studies, in which small-study effects and publication bias are expected to appear if present2. The separate analyses of test performance by Begg and Mazumdar and by Sterne, Gavaghan and Egger both confirm that the statistical information for detecting asymmetry is concentrated in this region. This is a further reason why the axis recommendation has remained the methodological default across the subsequent two decades of guidance, including the chapter on reporting biases in the Cochrane Handbook, which names the standard error as the standard choice for review authors constructing these plots459.

Identifying the axis used in a published funnel plot

Because the choice of axis is easily overlooked once a plot has been finalised and formatted, checking a published funnel plot for this specific error takes only a few minutes and does not require the underlying analysis to be repeated. The first and simplest check is to read the axis label, since a plot correctly constructed according to the recommendation of Egger and Sterne will typically label its vertical axis “standard error,” “1/standard error” or occasionally “precision,” whereas a plot constructed on the less reliable alternative will usually label it “sample size” or “n”2. Where the axis is unlabelled or ambiguous, a second check is to examine the shape of the scatter: a plot based on the standard error will generally show a smoothly narrowing spread towards the top, whereas a plot based on sample size more often shows an irregular or stepped pattern, because sample size does not track precision as smoothly as the standard error does for most effect measures.

A third, more diagnostic check is to look for pseudo-95% confidence interval lines on the plot. Because these diagonal lines are mathematically meaningful only on a standard-error axis, their presence is reasonably strong confirmation that the axis was specified correctly, whereas their absence is inconclusive rather than disqualifying, since some otherwise correctly constructed plots omit this optional feature3. Readers evaluating a meta-analysis for a journal, a grant panel or their own research synthesis can apply all three checks in sequence (label, shape and reference lines) before deciding how much weight to give any claim of asymmetry that the plot is used to support.

These checks should be applied even to plots from apparently reputable sources, since axis mislabelling tends to survive peer review precisely because it is an inconspicuous error rather than an obviously incorrect number. The 2000 analysis of funnel plot misuse by Tang and Liu found that technical construction errors of this general kind were sufficiently common in the published literature to warrant a dedicated methodological warning, separate from the more widely discussed question of what asymmetry signifies once it has been correctly plotted8. A reader who applies the three checks above before interpreting the shape of a plot is protected against precisely the failure documented in that analysis. Incorporating this brief verification into the way a research team reads any meta-analysis, whether its own or that of others, requires little time relative to the risk of accepting a conclusion about bias that the underlying chart was never constructed to support.

Key points

Key points

Standard error, not sample size, is the recommended vertical axis for a funnel plot because it has a consistent mathematical relationship to the effect estimate across different effect measures, which gives the plot a predictable shape under no bias and supports confidence-interval reference lines that a sample-size axis cannot.

Practical application

Practical application

  1. The vertical axis label should be confirmed before a published funnel plot is relied upon. The axis should show standard error, or an equivalent precision measure such as inverse variance, rather than raw participant counts.
  2. Software defaults that plot standard error automatically should be used where available. Most current meta-analysis software adopts this default specifically because of the 2001 Egger and Sterne guidance.
  3. Pseudo-95% confidence interval lines should be added when a funnel plot is constructed. These lines are meaningful only on a standard-error axis and provide readers with an immediate visual benchmark for judging the position of individual studies.

Frequently asked questions

Egger and Sterne concluded that it should generally not be used as the primary axis, since its relationship to the effect estimate is not sufficiently consistent to produce a predictable, interpretable shape, although sample size may still be reported alongside a plot for additional context.

It affects both. Egger’s test and related regression-based tests are based on the same standard-error relationship that makes the visual plot interpretable, so an inconsistent or incorrect precision measure undermines the formal test as well as the chart.

Egger and Sterne found that the mismatch between sample size and true precision is particularly pronounced for ratio measures such as odds ratios, for which the standard error of a study depends on the number of events observed and not only on the number of participants enrolled.

References

  1. 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
  2. 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
  3. 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
  4. 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
  5. 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
  6. 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
  7. 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
  8. Tang JL, Liu JLY. Misleading funnel plot for detection of bias in meta-analysis. Journal of Clinical Epidemiology. 2000;53(5):477-484. read 10 October 2026. View on Journal of Clinical Epidemiology
  9. 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
  10. Simmonds M. Quantifying the risk of error when interpreting funnel plots. Systematic Reviews. 2015;4:24. read 10 October 2026. View on Systematic Reviews

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). Why Is Standard Error Preferred Over Total Sample Size on the Vertical Axis of a Funnel Plot?. Pubrica. https://pubrica.com/academy/standard-error-in-funnel-plots/

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 Plot Axis Verification Checklist

Use this to check the vertical-axis choice before trusting or publishing a funnel plot.

[ ] Confirmed the vertical axis label is standard error (or an equivalent precision measure)
[ ] Flagged any plot using raw sample size as the vertical axis for closer scrutiny
[ ] Checked that the same effect measure was used consistently across all plotted studies
[ ] Added pseudo-95% confidence interval lines if using a standard-error axis
[ ] Verified software defaults match current methodological guidance before relying on them

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 every funnel plot with standard error as the default precision measure and checks third-party or legacy analyses for axis errors before they are relied upon in a submission.

Find something else

Search the Pubrica Academy

Access is free and requires no account. Enter a topic, or open the full library to filter by category, sub-topic or resource type.

Funnel plots that represent precision as reviewers expect

A scoping assessment is free and takes two business days. It provides a written assessment of what the work requires and which specialist would undertake it.

Get a free quote

Regulations

Check your project against this rule

  1. 1Your work
  2. 2Your details
  3. 3Send
What do you need?

Material you send is held under signed confidentiality, with access limited to the assigned team and encrypted transfer. It is not used to train anything and is not passed to third-party tools. Ask for the agreement before you send anything you would rather cover first.