Assessing the Global and Local Uncertainty of Scientific Evidence in the Presence of Model Misspecification

RT @Ohkubo2021: A paper proposes a bootstrap method to assess model selection uncertainty. Unlike EIC, target is ΔxIC, the diff. score. It…

This paper is also part of their "evidentialism" statistics, which, they arugue, extends likelihoodism approach of Royall(1997).

A paper proposes a bootstrap method to assess model selection uncertainty. Unlike EIC, target is ΔxIC, the diff. score. It aims to reflect the geometry of misspecification (relationship between null, alt., and true model) by assessing covariance of xIC. …

RT @FrontEcolEvol: New Research: Assessing the Global and Local Uncertainty of Scientific Evidence in the Presence of Model Misspecificatio…