|Image credit: Steve Ritter/Yang Ku/C&EN|
One proposed solution to publication bias, whereby studies with 'negative results' are not published, is to register (and peer review) protocols before the study is carried out.
A study by researchers at Cardiff University has explored how effective this approach is. The study found that protocols that were registered were more likely to report null findings, although the authors acknowledge that researchers may deliberately use this approach for hypotheses they anticipate will not produce positive findings.
The research was reported in Nature.