Confusing about running fixed effect variables?

ResoluteCarbon — Fri, 19 Feb 2021 21:48:39 GMT

Hi all experts,

I am new to SAS from STATA. I a just wondering that in STATA, if I want to do two-way fixed effects (for example for var1 and var2), Ineed to transform these variables to numeric type by using the code

encode var2, generate(var22)
encode var1, generate(var12)

This ocde can transform from the combination of words and numbers to numeric type that we can run regression afterwards (for example, they can transform string "111AQW" to numeric 111AQW).

I am not sure whether we need to transform all variables to numeric style when running regression for the categorized variables as the regression below

proc glm data=work.have;
    class var1 var2;
    model dependentvar var3 var4 var1 var2 /solution ss3;
run;
quit;

So, in this regression, I run dependentvar on var3 and var4 and control fixed effects for var1 and var2. So whther I need to transform var1 and var2 to numeric, and is there any code to do so ?

Thank you in advance.

Re: Confusing about running fixed effect variables?

StatDave — Fri, 19 Feb 2021 21:53:58 GMT

You don't. Variables in the CLASS statement can be numeric or character.

Re: Confusing about running fixed effect variables?

PGStats — Fri, 19 Feb 2021 22:43:34 GMT

as @StatDave said... and you must specify the model as

model dependentvar = var3 var4 var1 var2 / solution ss3;

topic Re: Confusing about running fixed effect variables? in Statistical Procedures

Confusing about running fixed effect variables?

Re: Confusing about running fixed effect variables?

Re: Confusing about running fixed effect variables?