Ho-Ju
10-10-2007, 10:34 PM
Are not the least squares version and the credibility version of the formula equal? If I recall from exam 4 correctly, it was shown they are equal. So why does Brosius say you should not use LS when changes are not random, but you should use the linear approx to Bayes when you get the same answer either way (assuming I'm right about them being the same)?