Difference between forms of factor coding: "*" and "!I"

Dear all,

I tried to fit a very simple model: yield ~ mu site !r variety. There were 100 varieties but coded discontinuously from 1 to 200. I fit the model using "variety *" first, then used "variety !I !SORT" for the same model. My expectation was that both would give the identical results, except that the BLUPs of non-existing varieties be set to zeros in the first model. But I don't understand why the two models actually differed in the fixed site effect, e.g., df, Wald F value, standard errors of BLUEs and predicted values. The model with "variety *" had much smaller standard errors of prediction. Could any one explain that to me? Thanks a lot!

Jason

-------------------- m2f --------------------

Sent using Mail2Forum (http://www.mail2forum.com).

Read this topic online here:
http://www.vsni.co.uk/forum/viewtopic.php?p=700#700

-------------------- m2f --------------------
Received on Tue Aug 03 2009 - 07:10:57 EST