Regression Model and Missing Values

[smajid]

Hi,

I plan to create a logistic regression model in SAS, but I am working with categorical data (independent variables) with missing values.

Should I eliminate the missing values before fitting the model? I'm leaning towards this because the goal is not to predict using missing categorical values.

OR

Should I create a new category for missing variables? If I do that, then I'd have to create an extra dummy variable.

Thank you in advance for your help!

SMajid

[Ksharp · Accepted solution]

If you have a big table, you could remove them.
OR using PROC MI to populate these missing value.

[Reeza]

Do you have random missing or systematic missing?

[ballardw]

By default the regression procedures will eliminate any record from analysis with missing values for any variable on the model statement.