P values adjusted for stopping rule
Probability of observing more extreme results under
the null hypothesis
Compute sampling distribution of test statistic under the
null
Requires a definition of “extreme” across analysis times
·Ordering of the outcome space