Funnel Plot Generator for Meta-Analysis
Create a funnel plot from your study results
Example with 13 studiesUse this funnel plot generator to plot study effect estimates against their standard errors. Enter estimates with standard errors or 95% confidence limits, or choose a supported summary-data format and let the calculator derive them. A plot appears with at least two valid, included studies. A small number of studies gives little basis for assessing asymmetry.
- Choose the format of your study data and the effect measure.
- Choose the model. Its pooled estimate is the center of the funnel.
- Check the funnel plot for asymmetry. With at least 10 studies, Egger's test follows below it.
How to generate your funnel plot
- Select Effect sizes and choose the effect measure that matches your studies.
- Select Standard error or 95% CI, then enter labels, estimates, and standard errors or confidence limits. Import or paste these values into the table above, then assign the corresponding columns.
- Select the pooling model. Its pooled estimate determines the center of the funnel guides.
- The Funnel plot appears below the heterogeneity results, followed by Egger's test when it is available. Both update when you change the included studies.
Which values should I enter?
Enter an effect estimate and its standard error, not the standard
deviation of participants' scores. All studies must use a comparable effect measure
and consistent direction. For odds ratios, risk ratios, or hazard ratios, choose
the corresponding (log input) option and enter the natural log ratio with its log-scale standard
error. If a study reports a 95% confidence interval instead, select 95% CI
and enter the estimate with its limits (log limits for ratios); the plotted standard error is
(upper - lower) / (2 * 1.96). This assumes a normal-approximation interval
that is symmetric on the analysis scale.
Alternatively, use Events & totals for two-group binary data, Means & SDs for two-group continuous outcomes, or Correlations for Pearson's r and sample size. The plot uses the derived estimates and standard errors. The meta-analysis calculator guide lists the required values and available effect measures for each format.
What does this funnel plot display?
| Plot element | Meaning |
|---|---|
| Each point | One included study estimate. |
| Horizontal axis | Effect on the analysis scale: a log scale for OR, RR, or HR and Fisher's z for correlations, with tick labels shown as ratios or r; the entered or derived effect scale for other measures. |
| Vertical axis | Standard error, reversed so more precise estimates appear nearer the top. |
| Vertical center line | Pooled effect under the selected model. |
| Dashed funnel limits | Pooled effect plus or minus 1.96 times the standard error at each height. These are pseudo 95% limits, not a confidence interval for the pooled effect. |
| Significance contours (optional) | Shaded bands around the null value where a single study would be significant at p < .10, p < .05, or p < .01. Turn them on under Show in the plot options. |
The funnel guides do not incorporate between-study variance and are not prediction intervals. Hartung-Knapp does not change these guides or Egger's test. Points outside them are not automatically erroneous or evidence of publication bias. Use the forest plot to inspect study confidence intervals and the heterogeneity calculator to assess variation between studies.
When is Egger's test available?
The calculator reports Egger's regression intercept, standard error, t statistic, degrees of freedom, and p-value only when at least ten included studies support an estimable regression. Fewer studies, or a regression that cannot be estimated, leave the test unavailable. Ten studies is a minimum here, not a guarantee of a reliable test.
Egger's test assesses small-study effects through funnel asymmetry. A small p-value does not establish publication bias, and a large p-value does not rule out missing evidence. The original regression can be misleading for odds ratios and standardized mean differences because the effect estimate and standard error can be related.
What can I conclude from asymmetry?
Asymmetry can reflect selective publication, differences in populations or study methods, heterogeneity, or chance. Interpret the plot alongside the search strategy, study quality, and research context. With significance contours: if studies seem to be missing mainly where results would be non-significant, selective publication is a more plausible cause; if they seem to be missing where results would be significant, other causes are more likely. This tool does not estimate how many studies are missing or perform trim-and-fill or selection models.
Keep one independent estimate per row. Use the table filter to exclude a study without deleting its values, and document the reason for exclusions. Use export / import above the table to save or reopen your input data. Method reference: Cochrane Handbook, Chapter 13 discusses funnel plots, asymmetry tests, and missing evidence.