Publication Bias

A study with strong or positive results gets shared far more often than one with weak or negative results. That imbalance makes an idea or treatment look like it works better than it actually does.

This bias systematically overrepresents statistically significant or favorable study outcomes in the published literature, relative to all the research that was actually conducted. The evidence base ends up skewed and the apparent effect size inflated, simply because of which results ever make it through editorial and dissemination processes.

A pharmaceutical company funds 12 clinical trials testing a new depression medication; 11 show modest or no improvement over placebo, but one shows strong positive results. Only that one successful trial gets published in a major medical journal, while the negative results gather dust in filing cabinets — leaving doctors and patients to overestimate the drug's effectiveness based on the incomplete picture the literature actually shows.

Researchers put more effort into writing up surprising results, leaving the weak ones unpublished. Editors and reviewers favor papers that look more interesting too, reinforcing the same imbalance from the other side.

Editorial policy, citation-driven prestige, and reviewer preference all feed into a selection process that privileges the significant outcome over the null one. That imbalance constrains which manuscripts actually pass through peer review, and the citations that follow only amplify the favored result further.

Encourage researchers to share every study result, even the null ones. Journals and funders that require or reward full reporting keep the record honest.

Mandatory pre-registration and registered reports tie publication to the methodology rather than the outcome, so null results still make it into the record. Adjusting editorial and funding incentives reduces the outcome-dependent selection driving the bias in the first place.

Overstated effect sizes; Wasted replication resources; Skewed policy decisions