[R-sig-ME] mixed effects logistic regression diagnostics

[Steven J. Pierce]

Stella,

Try looking into the capabilities of the influence.ME package and perhaps
also look at the articles cited in the help for that package. I haven't
reviewed it extensively, but it may have some of what you want. 

Steven J. Pierce, Ph.D. 
Associate Director 
Center for Statistical Training & Consulting (CSTAT) 
Michigan State University 
178 Giltner Hall 
East Lansing, MI 48824 
E-mail: pierces1 at msu.edu 
Web: http://www.cstat.msu.edu 

-----Original Message-----
From: S Mazeri [mailto:S.Mazeri at sms.ed.ac.uk] 
Sent: Friday, October 21, 2011 8:22 AM
To: Darren Norris
Cc: r-sig-mixed-models at r-project.org
Subject: Re: [R-sig-ME] mixed effects logistic regression diagnostics

Hi Darren,

Thank you for your reply.
My model looks like this:
m2 <-glmer(result~age+pastnt+buffevr+sex+ethnic2+(1|hcode),  
data=camqfever, family=binomial)
summary(m2)

result = outcome of the test
age+pastnt etc = fixed.effects
1|hcode = my random effects variable in this case its different herds

So I am not sure what and how to do diagnostics on it. Any ideas?

Thank you very much,

Stella


Quoting Darren Norris <doon75 at hotmail.com> on Fri, 21 Oct 2011 09:02:19
-0300:

> Hi Stella,
> Not sure what you are looking for with "cooks distance" etc
>
> Are you able to provide example of the syntax you are using to specify
> your model? As this could help to get some more specific guidance?
> There is a "profile" method for model diagnostics but I'm not sure if
> this is available for the binomial family or which versions of lme4
> this is currently available in? (The developers work so fast I can
> never keep up).
> Chapter 1 of the draft version of the new book by Doug Bates (
> http://lme4.r-forge.r-project.org/book/ ) has some useful information
> on model diagnostics .......
>
> hth,
> Darren



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.