Hi.
I want to evaluate collinearity in a dataset with multiple imputations. The dataset have 12 variables in 800 patients, one of them has 134 missing value.
I know that Stata would have automatically flagged and dropped highly collinear variable from my model.
I want to review this topic from a point of academic view.
With .collin x1, x2, x3... (obs=666) the VIF and tolerance values are very different than if run .logistic Y x1, x2, .... (obs=2006), and after use .vif, uncentered
With the command .mi estimate: logistic Y X1, X2, X3...., and then .vif , Stata issued an error message "nor appropiate after regress, nocons; use option incentered to get uncentered VIFs"
How can I evaluate the collinearity of my variables?
Thanks in advance,
I want to evaluate collinearity in a dataset with multiple imputations. The dataset have 12 variables in 800 patients, one of them has 134 missing value.
I know that Stata would have automatically flagged and dropped highly collinear variable from my model.
I want to review this topic from a point of academic view.
With .collin x1, x2, x3... (obs=666) the VIF and tolerance values are very different than if run .logistic Y x1, x2, .... (obs=2006), and after use .vif, uncentered
With the command .mi estimate: logistic Y X1, X2, X3...., and then .vif , Stata issued an error message "nor appropiate after regress, nocons; use option incentered to get uncentered VIFs"
How can I evaluate the collinearity of my variables?
Thanks in advance,
