Hi folks, I looking for a proof that shows that in multiple linear regression, R^2 does not change when a linear transformation is performed on the dependent variable and/or independent variables. It's pretty trivial for simple linear regression since R^2 is simply the square of the correlation (r) between x and y, and r is invariant under linear transformations on x and y. It's a little tougher for multiple regression... I tried expressing R^2 in its matrix representation (ala Draper and Smith pg. 91) but didn't get far. Any suggestions?
Jason Owen Math and CS University of Richmond . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
