Power Analysis for binary outcome

Deanna_Payne — Wed, 24 Nov 2021 06:08:30 GMT

Can someone please direct me to materials to help me calculate power analysis for a data set that has a binary outcome and binary/categorical variables? I know how to do via SAS for continuous variables but don't know how to do it for binary and can't find helpful forum answers. How would I define observed difference or standard deviation for these type of questions?

A basic null hypothesis for this research question would be: There is no difference in smoking abstinence between gender or race.

I.E.

My binary outcome is:

Smoking abstinence >=3months (0=no, 1=yes)

A binary/categorical variables would be:

gender (0=female, 1=male)

race (0=Black 1=AmericanNative 2=Hispanic 3=White)

An example of a continuous variable power analysis I can do:

proc power;
twosamplemeans test=diff
gmeans = 0 | 1.9
stddev = 4.8
npergroup = .
power = .8
alpha = 0.05;
run;

Re: Power Analysis for binary outcome

Ksharp — Wed, 24 Nov 2021 12:52:26 GMT

Check Rick's comment.

https://communities.sas.com/t5/SAS-Procedures/Superiority-testing-and-Proc-T-test/m-p/442740#M69303

topic Re: Power Analysis for binary outcome in New SAS User

Power Analysis for binary outcome

Re: Power Analysis for binary outcome