98
Stochastic Curtailment Comments
Neither conditional power nor predictive power
have good foundational motivation
•Frequentists
should use Neyman-Pearson paradigm and consider optimal
unconditional power across alternatives
•Bayesians
should use posterior distributions for decisions