Does anybody have any ideas why in a Cox regression (following multiple imputation) using a continuous variable as one of the covariates, it would drop some obs? I have identified that it is my one continuous variable that is the culprit. When I run the univariable model I get a model that appears to use all 909 obs, simlarly for my multivariable model using my categorical covariates, but when I add in my continuous variable it drops the obs down to 868. I should say that my exposure and outcome have no missing data so are not imputed, but all of my covariates are imputed. Any ideas? Thanks.
-
Login or Register
- Log in with
Comment