analyse of deviance

["Marc Choisy" <Marc.CHOISY@mpl.ird.fr>]

Hi everyone,

 

I am a little bit puzzled about specifying the test of an analysis of deviance after a glm. After a glm with a poisson family, what test should be used for an analysis of deviance? "Chi" or "F"? Some people use "Chi" and others "F" but I do not know what justify the use of "Chi"  and the use of "F". Same question when using quasi for the faily of the glm. Thanks for help.

 

Marc.

 

 

Marc Choisy
CEPM UMR CNRS-IRD 9926
IRD, 911 Avenue Agropolis BP 5045
34032 Montpellier Cedex 1, FRANCE
Tel : +33 (0) 4 67 41 61 80
Fax : +33 (0) 4 67 41 62 99
e-mail : choisy@mpl.ird.fr
web : http://viradium.mpl.ird.fr/cepm/index.htm