topic Dummy variable in Statistical Procedures

Dummy variable

yael — Thu, 20 Apr 2017 11:02:45 GMT

Hello

I am asking for your advice,

I have 3 dummy variables with 1/0. These are called :Afirm ,Bfirm ,Cfirm.

I also have a variable called FIRMS that have 3 options:1,2,3. When Afirm=1 than FIRMS=1, When bfirm=1 than FIRMS=2, When cfirm=1 than FIRMS=3.

If I run a regression, I understand the meaning of the coeffient of Afirm ,Bfirm ,Cfirm. But is it any meaning for the variable FIRMS?

What is preferable to add in the regression in order to get the difference between the firms?

Thanks for your help!

Re: Dummy variable

Rick_SAS — Thu, 20 Apr 2017 14:45:46 GMT

It sounds like FIRMS tells you all you need to know. Just use a CLASS statement in a regression procedure, like this:

CLASS FIRMS;

MODEL Y = FIRMS ...;

It is not necessary to use the dummy variables in regression. If you are using a SAS procedure (such as PROC REG) that does not support the CLASS statement, you can generate a design matrix (which will have either 3 or 4 columns) by following the advice in the article "Four ways to create a design matrix in SAS."