tpall/SRP: Calculate stable retrospective power

The standard use of null hypothesis significance testing (NHST) in controlling the long run quality of an experimental system works poorly in the confines of a single experiment. However, in the special case of massively parallel hypothesis testing, it is possible to convert it into an effective tool for describing the quality of a single experiment. To his end we introduce the concept of stable retrospective power (SRP), which complements the false discovery rate (FDR) and allows to reach meaningful conclusions about the quality of a single study. This enables to objectively plan for replication studies, and to estimate the quality of important parts of scientific literature without having to replicate them.

install.packages("remotes")
remotes::install_github("tpall/SRP")