topic Re: Rsqure in a fixed effect model in Statistical Procedures

Rsqure in a fixed effect model

Zak — Mon, 01 Mar 2010 12:29:22 GMT

Hi everyone!

My regression looks like this:

proc glm data=mydata;
absorb isin;
class year;
model tq = year
a b c /solution noint;
run;quit;

Among others, SAS gives me the Rsquare for the regression, which is really high. That is, as far as I understand it, because of the fixed effects I am using here. How could I exclude the impact of my fixed effects on the Rsquare?

I appreciate any help.

Cheers, Zak

Re: Rsqure in a fixed effect model

Paige — Mon, 01 Mar 2010 14:26:09 GMT

The usual test is to see if the enhanced model (with d in the equation) performs better than the first model (with only a b c in the equation) via the extra sum of squares principle (http://www.jerrydallal.com/LHSP/extra.htm).

Re: Rsqure in a fixed effect model

Zak — Mon, 01 Mar 2010 18:03:17 GMT

Thank you so much for your answer (which is the answer to my question I mistakenly posted in the other forum, I think). Anyway, that really helped me, although as far as I understand, a log likelihood ratio test is the test to utilize for the logistic regression which can be done by hand with the output provided by SAS.

Cheers, Zak