|
Activity Number:
|
353
- Research and Educational Tools
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 12, 2021 : 10:00 AM to 11:50 AM
|
|
Sponsor:
|
Section on Statistics and Data Science Education
|
|
Abstract #318559
|
|
|
Title:
|
The War on Statistical Significance: The American Statistician vs. The New England Journal of Medicine
|
|
Author(s):
|
Donald B. Macnaughton*
|
|
Companies:
|
MatStat Research Consulting
|
|
Keywords:
|
p-value;
hypothesis test;
false-positive error;
false-negative error;
replication crisis
|
|
Abstract:
|
This paper summarizes the main argument in a new book about the usefulness of statistical significance and p-values in the publication of scientific journals. The book frames the discussion in terms of a recent disagreement between editorials in two scientific journals. The book argues that a journal's use of a sensibly chosen threshold p-value (e.g., 0.05 or 0.01) as a necessary (though not sufficient) gateway to publication of a paper in the journal helps journal editors and researchers maximize the overall scientific and social benefit of scientific research. The gateway does this by producing a roughly optimal trade-off between the rates of costly false-positive and false-negative errors the journal makes in accepting or rejecting papers for consideration for publication.
|
Authors who are presenting talks have a * after their name.
