deviance and centering
Posted: Thu Sep 30, 2010 9:36 am
Hi all,
I have a question concerning deviance and centering of independent variables. I recentered the independent variable age for the whole sample for three separate multilevel models with Grade 7-9 students to a low, an average and a high value (to mean ages at Grades 7, 8, and 9, respectively). With each centering, the model deviances were different, and the models were, apart from centering, identical. Is this possible result, without it being a some kind of estimation problem? Can anyone give insight into what is going on? Thanks in advance.
Best wishes,
Antti
I have a question concerning deviance and centering of independent variables. I recentered the independent variable age for the whole sample for three separate multilevel models with Grade 7-9 students to a low, an average and a high value (to mean ages at Grades 7, 8, and 9, respectively). With each centering, the model deviances were different, and the models were, apart from centering, identical. Is this possible result, without it being a some kind of estimation problem? Can anyone give insight into what is going on? Thanks in advance.
Best wishes,
Antti