Large sample size but lots of missing data for ONE co-variate

#1
I am running a logistic regression analysis and I have pretty much have complete data for all variables in my model except for one important co-variate where just over 30% of data are missing! When I've run the model with the one variable with all the missing data, the same variables are statistically significant as the model without that one variable. My sample size is large (about 30,000) but it is reduced to about 21,000 with the inclusion of that one variable. Thoughts on the best way to address this?
 

noetsi

Fortran must die
#3
You probably need to either remove the variable or replace the missing data through multiple imputations (have fun with that). :(