Hi everybody,
I am new on STATA and maybe this can sound quite stupid, but I really need some help so I will ask anyway.
I'm running an epidemiological study on factors associated with drug resistance and I want to build a model to analyse how each of them actually linked to resistance with logistic regression.
I have ten variables and all of them are dichotomous.
I used stepwise estimation --> logistic --> p<0.1 for removal from the models. Is the command the correct one? It choses the variables which best fit the models by itself? I'm asking this because I repeated the command to check and actually when I enter the variables in a different order results and OR are different.. maybe I'm doing something wrong
Thank you for your help
0 Response to Choosing variables for logistic regression
Post a Comment