Dear All
In September 2001, I posted the message below concerning under dispersion in Poisson GLMs. A number of you kindly replied but you must have replied directly to me and not to the list as I can only find my original message in the archives. If any one can remember that far back, or has any further suggestions, I would be grateful.
Many thanks in advance
Guy
*---ORIGINAL MESSAGE FROM 2001
Dear All,
I'm carrying out some poisson GLMs on the number of eggs from starling
nests and want to know if there are any significant differences in the
mean number of eggs each year.
Ideally in poisson models one would want the residual deviance to be
close to the residual df indicating correct specification of the Poisson
model.
Usually in biological situations the residual deviance is much greater
than the residual df and this can be got round by estimating a dispersal
parameter from the data.
However, in this case I have the residual deviance/df ratio of 0.148
This affects the significance of terms in my final model as error bars
on the predicted means are considerably smaller
Could any one offer any advice on which to use, is 'under dispersion' a
valid concept?
Many thanks in advance
Guy
--------------------------------------------------------
Guy J Forrester
Biometrician
Manaaki Whenua - Landcare Research
PO Box 69, Lincoln, New Zealand.
Tel. +64 3 325 6701 x3738
Fax +64 3 325 2418
E-mail [log in to unmask]
www.LandcareResearch.co.nz
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
WARNING: This email and any attachments may be confidential and/or
privileged. They are intended for the addressee only and are not to be read,
used, copied or disseminated by anyone receiving them in error. If you are
not the intended recipient, please notify the sender by return email and
delete this message and any attachments.
The views expressed in this email are those of the sender and do not
necessarily reflect the official views of Landcare Research.
Landcare Research
http://www.landcareresearch.co.nz
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|