The drug did not work. The association was not there. The intervention changed nothing. Most researchers have a study like this in a drawer, never written up, because "who would publish it?" That instinct is understandable — and it is one of the most damaging habits in science. When null results stay unpublished, the literature slowly fills with only the positive half of reality, and everyone who later runs a meta-analysis inherits the distortion.
What publication bias does
If ten teams test the same hypothesis and one finds an effect by chance, publishing only that one makes a non-existent effect look real. Evidence syntheses then over-estimate the benefit, clinicians act on it, and other groups waste years pursuing a dead end that others already found. The cost is paid in duplicated effort and, in medicine, sometimes in patients.
Null is not the same as failed
Two things get confused:
- A null result is an answer: a soundly designed study that found no effect where one might have been expected.
- A failed study cannot answer its question — the design, sample or conduct prevents any conclusion.
The first is publishable and useful. The second needs fixing, not publishing. Your job in the manuscript is to convince the reader you are firmly in the first category.
How to write up a null result convincingly
- Show the study could have found an effect. Report your power calculation and the effect size the study was powered to detect. A null finding from an underpowered study says little.
- Give confidence intervals, not just p-values. "No difference (95% CI −0.4 to +0.5)" tells the reader you excluded a meaningful effect; a bare "p = 0.42" does not. See our guide to statistical reporting.
- Demonstrate the method worked. Positive controls, successful recruitment, and validated instruments show the null is about the hypothesis, not the execution.
- Interpret carefully. Absence of evidence is not evidence of absence — say what your data can and cannot rule out.
- Be honest about limitations. A frank limitations section strengthens a null paper more than any other kind.
Frame the contribution in the title and abstract. "No association between X and Y in a prospective cohort of 1,200 adults" is a finding, stated plainly — not an apology.
Where null results fit best
Journals that assess manuscripts on methodological soundness rather than perceived novelty are the natural home, as are journals and sections that explicitly welcome negative findings. Registered Reports are the strongest option of all: because the study is accepted on the strength of its design before the results exist, a null outcome is published exactly as a positive one would be.
The professional case
Publishing a null result protects your own record too. It establishes what you did, prevents others repeating it blindly, and — with preregistration — demonstrates that you reported what you set out to test rather than only what happened to work. That is a reputation worth more than one extra positive finding.