But in the case of the multiple regression Reu = ln(c/L) on (ln(c/ e),ln(c/u)), the PERL Module "Statistics::Regression" returns the standard errors of the coefficients (including intercepts), but not their variances nor covariancea. And it will not be trivial for me to try and get these on my own by messing with the internals of the module.
So, for the purposes of the possible "test" we've been discussing in my post of 823@957 and your post of 824@823, would standard errors of the coefficients suffice for our multiple linear regression?
That is, would you be able to construct a satisfactory test using slope variance, intercept variance, and (slope,intercept) covariance for our two simple linear regressions, but just the SE's of the coefficients (including intercept) for our multiple linear regression?
Thanks again as always for considering this question.