Why selective publication of statistically significant results can be effective

Why selective publication of statistically significant results can be effective

De Winter, J.C.F.
Happee, R.

Mechanical, Maritime and Materials Engineering

Biomechanical Engineering

2013-06-20

Concerns exist within the medical and psychological sciences that many published research findings are not replicable. Guidelines accordingly recommend that the file drawer effect should be eliminated and that statistical significance should not be a criterion in the decision to submit and publish scientific results. By means of a simulation study, we show that selectively publishing effects that differ significantly from the cumulative meta-analytic effect evokes the Proteus phenomenon of poorly replicable and alternating findings. However, the simulation also shows that the selective publication approach yields a scientific record that is content rich as compared to publishing everything, in the sense that fewer publications are needed for obtaining an accurate meta-analytic estimation of the true effect. We conclude that, under the assumption of self-correcting science, the file drawer effect can be beneficial for the scientific collective.

http://resolver.tudelft.nl/uuid:3b46f593-9d00-47ba-983d-d3af5cc18e93

https://doi.org/10.1371/journal.pone.0066463

Public Library of Science

1932-6203

Plos One, 8 (6), 2013

Institutional Repository

journal article

(c) 2013 de Winter, Happee. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.