|
Activity Number:
|
245
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 1, 2016 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #321238
|
|
|
Title:
|
Massive Significance Testing in Genetic Epidemiology Amid Replicability Crisis
|
|
Author(s):
|
Dmitri Zaykin* and Olga A. Vsevolozhskaya
|
|
Companies:
|
National Institute of Environmental Health Sciences and University of Kentucky
|
|
Keywords:
|
reproducibility ;
hypothesis testing ;
genome-wide ;
multiple testing ;
false discovery rate ;
genetic associations
|
|
Abstract:
|
Recent flurry of concerns about replicability and reproducibility has been igniting heated discussions of whether or not science is in crisis. Specific worries over the statistical crisis in science are largely attributed to misapplication of classical approaches like that of the null-hypothesis significance testing (NHST). Shortcomings of NHST are thought to be prominent in multiple testing settings, such as in human genetics research, where the number of statistical tests can reach tens of millions. We examine a number of recent publications critical of NHST and P-values and find that the criticism is at times misguided. We accentuate some redeeming features of P-values and study relations of the smallest P-values from genome-wide association experiments to estimated Bayesian probabilities of hypotheses.
|
Authors who are presenting talks have a * after their name.
