A two-step manuscript submission process can reduce publication bias

Abstract

Much of what is researched is never published. This would not be of great concern if the selection of what we read would occur irrespective of study outcomes. Unfortunately, the reverse is the case: "positive" studies have a much larger chance of acceptance after editorial and peer review than "negative" ones. Several solutions to this problem of publication bias have been discussed or implemented, but none seem to be very effective. In this article, the approach of implementing an editorial and peer-review procedure that is blinded to study outcomes is discussed. This would require a two-step submission procedure of manuscripts: first a version including just the introduction and methods and in some cases followed by a second submission including results and discussion. The pros and cons of such an approach are discussed.

Keywords: Bias; Blinding; Editorial review; Peer review; Publication bias; Statistics.