February, 2003
© 2000, 2001, 2003 Scott S. Emerson, M.D., Ph.D.
Because the estimate of the treatment effect is no longer
normally distributed in the presence of a stopping rule, the
frequentist inference typically reported by statistical
software is no longer valid
The standardization to a Z statistic does not
produce a standard normal
·The number 1.96 is now irrelevant
·
Converting that Z statistic to a fixed sample P
value does not produce a uniform random variable under the null
·We
cannot compare that fixed sample P value to 0.025