1. I tried numerous software packages, including PASS, but it appears that they can only be used when the model contains a single variable. So I was wondering if there was a function (mostly in R) or software that could be used when there are additional variables in the model. 2. Also, if the value of correlation between the variable of interest and the other covariates in the model is unknown, is there any method to compute the sample size.
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
While I am not 100% sure, it is very unlikely that there is an analytic solution for either of these problems. The various power software programs are usually very quick to add capabilities when they become known.
So, you will probably have to simulate the data and do power analysis that way.
-
$\begingroup$ We need to know the goal of the analysis. Is it to estimate the adjusted effect of a single variable, or to get a reliable overall model? $\endgroup$ Commented Jun 18 at 12:32
-
$\begingroup$ @ Frank Harrell, Dear Professor Harrell, thank you so much for your remark. The purpose is to estimate the adjusted effect of a single variable. The model includes a variable of interest (binary) and four other covariates. However, I would be grateful if you could please supply the solution for both scenarios (estimate the adjusted effect of a single variable and also obtaining a reliable overall model). $\endgroup$– Stat2024Commented Jun 18 at 14:25