50
Operating Characteristics
Generally the same for all stopping rule s
–Frequentist
power curve
•Type
I error (null) and power (design alternative)
–Sample
size requirements
•Maximum,
average, median, other quantiles
•Stopping
probabilities
–Inference
at study termination (at each boundary)
•Frequentist
inference
•Bayesian
inference under spectrum of priors
–Futility
measures
•Conditional
power, predictive power