Skip to content
Prev 258435 / 398502 Next

Questions about lrm, validate, pentrace

(11/04/29 22:09), Frank Harrell wrote:
Thank you for your comment. I am able to be confident about my model now.

The difference you saw is caused
A lot of previous studies have demonstrated that poor outcome is more 
frequent in treat2 than in treat 1. So, I coded treat1 as 0, and treat2 
as 1 in the first mail. Then, I came back to the original coding of 
treat1 and treat2 in the newer mail. According to your answer, I guess I 
am OK. :-)

Prof Harrell, Your book (Rregression Modeling Strategies) and many kind 
comments helped me a lot. Thank you very much again.

--
KH