RT Journal
A1 Janosky JE
T1 THe ethics of underpowered clinical trials
JF JAMA
JO JAMA
YR 2002
FD November 6
VO 288
IS 17
SP 2118
OP 2119
DO 10.1001/jama.288.17.2118
UL http://dx.doi.org/10.1001/jama.288.17.2118
AB
I believe that the concept of statistical power as an absolute is misleading.
Statistical power is typically presented as a dichotomy that defines a cutoff
of .80 or greater. In fact, the concept of statistical power exists along
a continuum. A value for statistical power is intrinsically related to the
level of statistical significance, the level of treatment differences, variability
of response, research question of interest, and sample size. Separating statistical
power from these other parameters misinterprets the issue of statistical power.
A clinical trial has a certain level of statistical power associated with
it, given the level of these other parameters.