What is it about?
This paper proposes a way to correct estimates or other decisions for uncertainty that is not quantified by the statistical model. The practical example is an application of the correction to null hypothesis significance testing.
Featured Image
Why is it important?
No statistical model can quantify all of the uncertainty present in data analysis. As a result, statistical methods tend to underreport uncertainty in the conclusions, possibly leading to many of the false positives associated with the replication crisis. This paper provides a simple way to correct the results of null hypothesis significance testing for unquantified uncertainty.
Perspectives
For further reading, follow the "Related papers" link.
David R. Bickel
University of North Carolina at Greensboro
Read the Original
This page is a summary of: Departing from Bayesian inference toward minimaxity to the extent that the posterior distribution is unreliable, Statistics & Probability Letters, September 2020, Elsevier,
DOI: 10.1016/j.spl.2020.108802.
You can read the full text:
Resources
Contributors
The following have contributed to this page
