[RsR] calculating deviance for glmrob

[Andreas Ruckstuhl]

Dear Alex

Yes, the deviance is deliberately not implemented. Hence, the statement 
in the help file is incorrect.

If you want to compare your model with another model, please use anova 
(anova.glmrob). However, you don't find any deviance calculations there 
as well but the differences of robust quasi-likelihoods. The reason is 
the difficulties with the interpretation of ("raw") robust 
quasi-likelihoods. - Do you have any particular applications of ("raw") 
robust quasi-likelihoods in mind?

Thank you very much for pointing out the inconsistency in the help file.

All the best,
Andreas




------- Ursprüngliche Nachricht --------
Von: Alex Holcombe <holcombea using me.com>
Gesendet: Tue, 26 Oct 2010 17:57:47 +1100
An: r-sig-robust using r-project.org
Betreff: [RsR] calculating deviance for glmrob

> Standard glm returns a deviance field, which is apparently minus twice the maximized log-likelihood.
> glmrob claims to do the same in its help page, but seems to always return NULL for deviance.
> 
> It seems to be an unimplemented feature.  Does anyone have code or an algorithm that can calculate it?
> 
> Any pointers appreciated
> 
> 
> Alex Holcombe
> www.psych.usyd.edu.au/staff/alexh/
> 
> _______________________________________________
> R-SIG-Robust using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-robust

-- 
----------------------------------------------------------------------

Prof. Dr. Andreas Ruckstuhl
ZHAW Zürcher Hochschule für Angewandte Wissenschaften
IDP Institut für Datenanalyse und Prozessdesign
Rosenstrasse 3                    Tel.  : +41 (0)58 934 78 12
Postfach                          Fax   : +41 (0)58 935 78 12
CH-8401 Winterthur                e-Mail: Andreas.Ruckstuhl using zhaw.ch
                                   WWW   : http://www.idp.zhaw.ch