On the SAS website, I found this info:

Fisher (1970, p. 199) describes the following practical applications
of the transformation:


    testing whether a population correlation is equal to a given
value

    testing for equality of two population correlations

    combining correlation estimates from different samples

Does it all apply to regression coefficient (r square) too?  Could I
apply the Fisher z transformation to rsq and then compare two
transformed rsq, or get a mean transformed rsq from different
samples?

I know proc corr has a new option called Fisher and transforms the
coefficient for you. In the past, I did it by using the formula. I
would use the formula to transform rsq, that would be no problem, if
the the above stated still applies for rsq.

Thanks!