It is well known that an Analysis of Variance Model can be set up as a General Linear Model or a Regression Model ......but I haven't seen a treatment for the following issue.
A colleague has asked me why one tests the significance of a main effect ...say there are three groups or classes ....simultaneously via an F test rather than simply specifying two dummy variables and a mean in a Regression Model and doing a step-down. I contend that you must first test the hypothesis by deleting both dummies and comparing the increase in the SOS divided by two versus the the MSE for a complete model including the two dummies ...and if the hypothesis of no effect is rejected one must then go forward with both dummy variables and not even review the individual t values for each of the two dummies. He contends that one can simply stepdown from the complete model and delete one-by-one any of the two dummy variables until parsimony is achieved. This avoids the portmanteau F test and relies strictly on the individual T values ... I can't find any textbook that supports my position and suggests that what he is doing is flawed. Thanks for your help. Dave Reilly Automatic Forecasting Systems 215-675-0652 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
