How to run multiple linear regression with (explanatory) categorical variables with >2 levels

[tpakhomova]

Hi, 

I would like to perform a multiple linear regression, and was wondering is PROC REG or PROC GLM/GLMSELECT are my better options when some of my explanatory variables are 3 and 4 levels? I have not used REG or GLM before, so am not familiar with either. I have found the explanation for dummy variables to be somewhat confusing, especially how to determine the reference level (is it always 0, or 1?). Looking for some advice on how to best proceed. 

[WarrenKuhfeld · Accepted solution]

Use glmselect. The documentation has a lot of information about coding.

Editor’s Note: This video showing how to build multiple linear regression models with and without interaction may be helpful as well.

[tpakhomova]

Do I need to create dummy variables for GLMSELECT? Or is that only for PROC REG?

[WarrenKuhfeld]

I would not have recommended glmselect if you had to do your own coding.