发帖
雷达卡
I would like to seek help with such a question: is there any way that we compare the relative magnitude of covariates upon the likelihood/odds of failure (e.g. recurrence of T.B.) in Cox model?
Best to my knowledge, it is technically feasible to compare the substantive impacts of IVs upon DV in logit regression. I have come across two types of situation. In the first situation, people place all main IVs in one model. In the other, for fear of high level collinearity or out of other consideration, scholars control for the same set of variables so that the only difference between models is the main IV. In both siutations, we can run the post-regression estimation and find out "the odds changes in response to 1 st. dev. increase in X."
But what can we do with the Cox model? For instance, I am going to assess the relative magnitude of two factors, namely meat intake and work intensity, upon the return of heart attack. Sorry for this very awkward example. I know nothing about medical science. I only want to give an example.
My stat tool is STATA.
Many thanks!
京公网安备 11010802022788号
