Choice of methods for inference
Fixed sample tests
·All frequentist methods described here agree with
each other
Group sequential tests
·No method is uniformly better
·Usually fairly good agreement between various methods
·Failure
to agree can be informative regarding time trends in data