Evidence-based peer review can make research more robust

Some tools, such as author blinding, seem to reduce retractions, while others are associated with more, say Serge Horbach and Willem Halffman.

Concerns about the reliability of the scientific literature have been sparked by biomedical researchers finding that a worrying proportion of published articles contain flaws, are not reproducible or are outright fraudulent. This may have several consequences, including sending research down fruitless avenues, wasting valuable resources, skewing meta-analyses, or building policy or treatments on insecure grounds. It may even harm public trust in science and threaten funding.