Date created: | Last Updated:

Category: Project

Description: Recently, a number of statistical reformers have argued for conceptualizing significance testing as analogous to diagnostic testing, with a "base rate" of true nulls and a test's error probabilities used to compute a "positive predictive value" or "false discovery rate". These quantities are used to critique statistical and scientific practice. We argue against this; these quantities are not relevant for evaluating statistical tests, they add to the confusion over significance testing, and they take the focus away from what matters: evidence.