Publication bias and the funnel plot
- Bias
Studies with positive results get published more than negative ones; when the negatives are missing, the effect looks bigger.
What it is
Publication bias happens when whether a study is published depends on the direction or significance of its results. Small studies with no effect go unpublished, and a review that finds only what was published overestimates the benefit. The funnel plot places each study by its effect and its precision: without bias it forms a symmetrical funnel; if small studies are missing on one side, it is asymmetrical. It is one form of bias due to missing results, alongside selective outcome reporting.
How it arises
- Authors and journals prefer significant, novel results.
- Small studies with null results are abandoned or published late.
- Within a published study, favourable outcomes are reported and others left out (selective reporting).
Worked example
- Data (made up): 20 trials of a treatment were registered: 10 with a positive effect and 10 with no effect.
- 9 of the 10 positive trials were published (90 %) and 3 of the 10 null trials (30 %).
- Published: 9 + 3 = 12 trials; positive ones make up 9 / 12 = 75 %.
- In reality, positive trials were 10 / 20 = 50 %: the published literature exaggerates the favourable share.
- Probability of publication: 90 % / 30 % = 3 times higher for a positive result.
- In the funnel, the bottom corner on the no-effect side is empty: the small null trials are missing.
- Reading: a meta-analysis of published trials alone would overestimate the benefit; the review must search the registries.
Markers
- Funnel
- Asymmetry
- Trial registry
- Selective reporting
How to prevent or control it
- Register every trial before it starts, in public registries brought together by the WHO International Clinical Trials Registry Platform.
- Search trial registries, conference abstracts and grey literature during a systematic review.
- Compare the outcomes in the registered protocol with those published to detect selective reporting.
Common pitfalls
- Blaming all funnel asymmetry on publication bias: heterogeneity, poorer small studies or chance also cause it.
- Interpreting a funnel with 5 studies: it lacks the power to show asymmetry.
Test yourself
Think of the answer, then tap the question to see it.
What is publication bias?
When whether a study is published depends on the direction or significance of its results, so the published evidence overestimates the effect.
What does an asymmetrical funnel plot show?
That small studies are missing on one side (usually the no-effect side); it suggests publication bias, though other causes are possible.
Which measure helps most to prevent publication bias?
Prospective registration of every trial in a public registry before the first patient is enrolled.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Chapter 13: Assessing risk of bias due to missing evidence in a meta-analysis (Cochrane, 2026); International Clinical Trials Registry Platform (ICTRP) (World Health Organization, 2026).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.