Increasing the Replicability for Linear Models via Adaptive Significance Levels
Methodology
2021-01-06 v3
Abstract
We put forward an adaptive alpha (Type I Error) that decreases as the information grows, for hypothesis tests in which nested linear models are compared. A less elaborate adaptation was already presented in \citet{PP2014} for comparing general i.i.d. models. In this article we present refined versions to compare nested linear models. This calibration may be interpreted as a Bayes-non-Bayes compromise, of a simple translations of a Bayes Factor on frequentist terms that leads to statistical consistency, and most importantly, it is a step towards statistics that promotes replicable scientific findings.
Cite
@article{arxiv.2003.06938,
title = {Increasing the Replicability for Linear Models via Adaptive Significance Levels},
author = {D. Vélez and M. E. Pérez and L. R. Pericchi},
journal= {arXiv preprint arXiv:2003.06938},
year = {2021}
}