February, 2003
© 2000, 2001, 2003 Scott S. Emerson, M.D., Ph.D.
Criteria for sample size (cont.)
Sometimes constrained by practical limitations
·Compute alternatives that can be discriminated
from the null
-Frequentist:
alternative for which study has desired power
-Bayesian: alternative for which posterior probability
is sufficiently high when null posterior probability is
low
·Determine power (frequentist) or posterior probability
(Bayesian) criterion which corresponds to a particular
alternative