Publication Bias

Publication bias happens when studies with strong or positive results get shared more often than studies with weak or negative results. This makes it look like an idea or treatment works better than it really does.

Publication bias is the systematic overrepresentation of statistically significant or favorable study outcomes in the published literature relative to all conducted research. This skews the evidence base and inflates apparent effect sizes by selectively filtering which results pass 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 1 shows strong positive results. The company publishes only that one successful trial in a major medical journal, while the negative results gather dust in filing cabinets, leading doctors and patients to overestimate the drug's effectiveness based on the incomplete picture they see in the literature.

Researchers do more work to write up surprising results while leaving weak results unpublished. Editors and reviewers also pick papers that look more interesting, so those get published more.

Within publication_visibility_economy_systems__weighting_asymmetry, editorial policies, citation-driven prestige, and reviewer preferences create a weighted selection mechanism that privileges significant outcomes. This asymmetry constrains which manuscripts traverse peer review and attain visibility, amplifying favored results through cumulative citations.

Encourage sharing of all study results, even when they show no effect. Journals and funders can require or reward full reporting.

Mandate study pre-registration and promote registered reports to align publication with methods rather than outcome; ensure data repositories host null results. Adjust editorial and funding incentives to reduce outcome-dependent selection and rebalance visibility.

Overstated effect sizes; Wasted replication resources; Skewed policy decisions