Testing and adjusting for publication bias

How can the scientific literature provide unbiased conclusions if it represents a biased sample of available studies? Publication bias refers to phenomena arising from bias in submitting, reviewing, accepting and publishing scientific results. Direct and indirect methods for investigating publication bias are now readily available, but indirect methods are generally open to alternative interpretations. Publication bias distorts attempts to review a scientific field quantitatively if the likelihood of locating completed studies depends on the strength or direction of the findings of quantitative studies. It is the responsibility of researchers, reviewers and editors to address issues of bias to ensure the existence of an unbiased literature.