Severity and Trustworthy Evidence: Foundational Problems versus Misuses of Frequentist Testing

Abstract For model-based frequentist statistics, based on a parametric statistical model ${{\cal M}_\theta }({\bf{x}})$ , the trustworthiness of ensuing evidence depends crucially (i) validity probabilistic assumptions comprising (ii) optimality inference procedures employed, and (iii) adequateness sample size ( n ) to learn from data by securing (i)–(ii). It is argued that criticism postdata severity evaluation testing results small Rochefort-Maranda (2020) meritless because it conflates [a] misuses with [b] genuine foundational problems. Interrogating this reveals several misconceptions about trustworthy estimation-based effect sizes, which are uncritically embraced replication crisis literature.