10-02-2013 07:13 AM
Does anybody know the formula for the Lack of Fit test in case of WEIGHTED regression (linear and nonlinear)? proc reg has for example the lackfit option (also for weighted LS). Thanks!
10-02-2013 10:48 PM
Educated guess: Since the lack of fit test is a F test, it will involve the ratio of sums of weighted squares instead of the ratio of sums of squares, with the same degrees of freedom.