When a study finds nothing and is never published, someone else runs the same study — spending the same resources to reach the same non-conclusion.
And the published literature comes to overstate every effect, because only the studies that found something appear in it.
Why null results matter
- They prevent repetition of unproductive approaches.
- They correct the overstatement caused by selective publication.
- They are essential to any evidence synthesis.
- A well-designed study that found nothing is still a finding.
- The distortion is cumulative and affects the whole field.
Why they go unpublished
- Perceived difficulty of publishing them.
- Assessment rewarding positive findings.
- Authors abandoning them for more promising work.
- Belief that reviewers will reject them.
- Some of this is real and some is assumed without testing.
Making a null result publishable
- The design must be sound enough to support the conclusion.
- Show the study could have detected an effect if present — this is the essential point.
- Report what was tested and how.
- Distinguish no effect detected from no effect exists.
- State implications for future work.
Where to publish
- Some venues explicitly welcome null results.
- Registered reports, where accepted before results are known.
- Preprints, which at least make the work findable.
- Include null findings within larger papers where appropriate.
- Repository deposit with a data description at minimum.
Framing
- Lead with what was tested, not with the absence.
- Emphasise the design quality.
- State clearly what this rules out.
- Note what remains possible.
- Avoid overstating in either direction.
What systems can do
- Recognise null results in assessment.
- Support registered report formats.
- Encourage preregistration.
- Treat evidence synthesis as valued work.
- Selective publication is a system problem, not an individual failing.
One thing worth remembering
A null result from a well-designed study is a finding — provided you can show the study could have detected an effect had one been there.
That demonstration is what separates a publishable null result from an underpowered study, and it is the part most often left out.
Câu hỏi thường gặp
Why do null results matter?
They prevent repetition of unproductive approaches, correct the overstatement caused by selective publication, and are essential to evidence synthesis.
Why do they go unpublished?
Perceived difficulty of publishing, assessment rewarding positive findings, authors abandoning them for more promising work, and assumed reviewer rejection.
What makes a null result publishable?
A sound design, and crucially showing the study could have detected an effect if present — while distinguishing no effect detected from no effect exists.
Where can null results be published?
Venues that explicitly welcome them, registered reports accepted before results are known, preprints, within larger papers, or at minimum repository deposit.
What separates a publishable null result from an underpowered study?
Demonstrating the study could have detected an effect had one been there — the part most often left out.
