R help - Oct 2010 - How do I set the dispersion parameter in poisson glm?

Dear R users,

I would like to fit a glm with Poisson distribution and log link with a known
dispersion parameter.  I do not want to estimate the dispersion parameter.  I
know what it is, so I simply want to fix it at a constant for this and other
models to follow.  My simple, no covariate model is:

Tall.glm<-glm(Seedling~1, 
	family=poisson, 
	offset(log(area)),
	data=tallPSME.df)

I want to fix the dispersion parameter at 2.5.  How can I do this, please?

Many thanks in advance for your help,

Manuela
 
>::<>::<>::<>::<>::<>::<>::<>::<>::<
Manuela Huso
Consulting Statistician
201H Richardson Hall
Department of Forest Ecosystems and Society
Oregon State University
Corvallis, OR   97331
ph: 541-737-6232
fx: 541-737-1393

On Thu, 7 Oct 2010, Huso, Manuela wrote:
> Dear R users,
>
> I would like to fit a glm with Poisson distribution and log link with a
known dispersion parameter.  I do not want to estimate the dispersion parameter.
I know what it is, so I simply want to fix it at a constant for this and other
models to follow.  My simple, no covariate model is:
>
> Tall.glm<-glm(Seedling~1,
> 	family=poisson,
> 	offset(log(area)),
> 	data=tallPSME.df)
>
> I want to fix the dispersion parameter at 2.5.  How can I do this, please?
The parameter estimation is invariant to the choice of dispersion 
parameter, hence you don't have to change anything in the glm() call 
above.

Only for the inference you have to set the dispersion, e.g.

summary(Tall.glm, dispersion = 2.5)

hth,
Z
> Many thanks in advance for your help,
>
> Manuela
>
>>
::<>::<>::<>::<>::<>::<>::<>::<>::<
> Manuela Huso
> Consulting Statistician
> 201H Richardson Hall
> Department of Forest Ecosystems and Society
> Oregon State University
> Corvallis, OR   97331
> ph: 541-737-6232
> fx: 541-737-1393
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

I was not using a package to calculate QAIC.  I will look into your suggestions,
though.  I thought it would be simple to ask glm to just fix the overdisperion
at a particular value.  Apparently it is not.

Thank you for all you help.

Manuela

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of Ben Bolker
Sent: Thursday, October 07, 2010 12:04 PM
To: r-help at stat.math.ethz.ch
Subject: Re: [R] How do I set the dispersion parameter in poisson glm?

Huso, Manuela <manuela.huso <at> oregonstate.edu> writes:
> 
> Thank you so much for your reply, Z.  Of course, you are correct about
>  the overdispersion not affecting
> parameter estimation.  But my goal is comparison of several models 
> using QAIC and the dispersion
> parameter enters the QAIC calculation.  I suppose I could calculate 
> the QAIC of all models using my fixed
> dispersion parameter, but I was hoping I could just set it in the 
> glm call and let the software calculate
> QAIC appropriately for me.
  [snip]

   Which QAIC() function are you using?
  The ones in the MuMIn,AICcmodavg, and bbmle packages all
have parameters that should allow you to set the dispersion
parameter when you calculate QAIC.

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.