Power calculation for proportions

[Cyril_FR]

Hello,

I want to ask why the following code is not working:

proc power;
TwoSampleFreq
test =pchi
groupproportions =(1 0.80)
nullproportiondiff =0
NPerGroup =20
SIDES =1
POWER =.
Alpha =0.05;
run;

invalid input
WARNING: The magnitude of the effect is ill-defined in one of the scenarios in the output table.

I think that it's due to my assumption that 100% of the patients in the placebo group develop the disease. Is it too strong?
Thanks in advance,
Best regards,
Cyril

[Doc_Duke]

You got it. Underlying any experimental design is the concept that there will be some variability in the process. If you are requiring a 100% effect (or 0%) in one group, then you have no variability in that group.

[Cyril_FR]

Thanks DOC