What is it about?
In previous analyses of acceptance rates of papers with positive and negative finders, it has been naively assumed that authors take no account of probability of acceptance when making submissions. If negative submitted papers are of higher quality than positive ones equal acceptance rates would be consistent with a bias in favour of positive studies. When studied carefully, there is some evidence that this is the case.
Featured Image
Photo by Carlos Muza on Unsplash
Why is it important?
If what is claimed here is true, the solution to bias in favour of positive papers does not rest with author but with editors.
Perspectives
My own interest from this is sparked by realising that although the evidence based medicine movement has laudably stressed the important of unbiased designs, for example randomised clinical trials, when assessing the effect of interventions, in "meta-research" of the field this has been often forgotten. It is necessary, when analysing observational data to consider the possible presence of "data filters". All too easily we make the implicit assumption that data arise as if they came from an RCT. Often this is not an appropriate analogy.
Professor Stephen J Senn
Consultant Statistician
Read the Original
This page is a summary of: Misunderstanding publication bias: editors are not blameless after all , F1000Research, December 2012, Faculty of 1000, Ltd.,
DOI: 10.3410/f1000research.1-59.v1.
You can read the full text:
Resources
Contributors
The following have contributed to this page
