What is it about?
Popular false discovery rate methods do not suffer from the excessive conservatism of conventional methods of multiple testing. However, they go too far in overcoming that conservatism, to the point of introducing an anti-conservative bias.
Featured Image
Why is it important?
The negative bias of popular false discovery rate methods means the actual probability of making a false discovery is higher than the rate indicated by the method. That can result in unintentionally exaggerated findings, contributing to the replication crisis. The practical effect of the bias can be huge, as you seen when playing with this software: https://zenodo.org/record/3234944 How to use the software: https://bit.ly/2Vbb1ln
Perspectives
The method of this paper is explained in Chapter 6 of D. R. Bickel (2019). Genomics Data Analysis: False Discovery Rates and Empirical Bayes Methods. Chapman and Hall/CRC, New York. https://davidbickel.com/genomics/
David R. Bickel
University of North Carolina at Greensboro
Read the Original
This page is a summary of: Correcting false discovery rates for their bias toward false positives, Communications in Statistics - Simulation and Computation, June 2019, Taylor & Francis,
DOI: 10.1080/03610918.2019.1630432.
You can read the full text:
Resources
Genomics Data Analysis: False Discovery Rates and Empirical Bayes Methods
Chapter 6 explains this paper in a simple way.
Interactive comparison of false discovery rates and local false discovery rates
This software gives an intuitive understanding of how popular false discovery rate methods lead to too many statistically significant findings.
How to use "Interactive comparison of false discovery rates and local false discovery rates" (Bickel, 2019)
Instructions for the software mentioned
Contributors
The following have contributed to this page
