Hi,

I am trying to impute two separate variables. When trying to determine an appropriate imputation model, I regress each of the two variables on my other analysis variables and other auxiliary predictors, and both variables seem to be strong predictors of the other.
I run into the problem cited here when using mi chained, in that I cannot include the variables in the imputation model given they will be imputed.
Code:
set more off
mi impute chained (poisson, include(var2)) var1 ///
(reg, include(var1)) var2  ///
= var 3 var4 var5 var6 augment  add(10) rseed(4583) force
Specifically, "var1: imputation variable cannot be also specified as independent variable".

I realize imputing separately and using the imputed values for each in the imputation model of the other is unadvisable, but is one approach here to impute separately and use the original values of the variables?

Thanks in advance.