Skip to content

Dispersion in summary.glm() with binomial & poisson link (fwd)

8 messages · Brian Ripley, Ben Bolker, richards@toutatis.pci.upmc.edu +4 more

Brian Ripley
# 
Yes, that was exactly my concern.  What gnlr does makes sense to me,
but a phi>1 binomial or Poisson only makes sense as a quasi model.  In
the distributions Jim has coded the variance/mean ratio is not
constant (as it is for the phi-Poisson), for example, and the implied
quasi-likelihood (where it exists) differs from those likelihoods.

But GLIM and S and R will in their own ways fit those models, and
I want to look hard to see if what they do in R is sensible (and
some things seem not to be, as Ben Bolker pointed out ca April 19).
Ben Bolker
# 
Thanks: I've been following this discussion with interest, since I had
wondered if my previous message had disappeared into the void ... I'm
getting lots of the information I wanted from this exchange.

   While we're on the topic: would it make any sense in this context to
implement the F-test suggested in MASS3 p. 215 (eq 7.10), or does the fact
that "this must be used with some caution in non-Gaussian cases" suggest
that it would be better to leave it out and not let people get into
trouble using it?

  Should the documentation about scale parameters say that if you're
heavily into estimating scale parameters you should probably be using
quasi-likelihoods instead?

  Just out of curiosity, is there documentation for profile.glm beyond
what's in MASS3?  (Or should it just be linked/added to the profile.nls
help?)  (I'm thinking about writing a profile.ms or something like that;
I've been having my students look at likelihood profiles a lot.)

  I will look into Jim Lindsey's code (which sounds like it does specific
extensions of binomial/Poisson rather than quasi-likelihood?)

 thanks all,
    Ben
richards@toutatis.pci.upmc.edu
# 
Hello:

I am trying to use the function

scatterplot3d(), obtained from:

    ##  scatterplot3d, 0.2.10 , 17.04.00,
    ##  Uwe Ligges <ligges at statistik.uni-dortmund.de>,
    ##      http://www.statistik.uni-dortmund.de/leute/ligges.htm


I had originally seen this function discussed here on the list, I think by
one of the authors.  Anyway, I need to know whether anyons has heard of
the function xyz.coords(), which the scatterplot3d function calls.  I get
an error message, under R 1.0.0 under NT4, as follows:

 couldn't find function "xyz.coords"
    
I am clueless.  Am I supposed to know about this function, as the autghors
seem to assume?  Thanks for any help.

				Tom Richards

:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:
: Thomas J. Richards              _/   _/  _/_/_/    _/_/  _/_/_/  :
: Statistician                    _/   _/  _/   _/ _/   _/   _/    :
: Biostatistics Center            _/   _/  _/   _/ _/        _/    :
: U. Pittsburgh Cancer Institute  _/   _/  _/_/_/  _/        _/    :
: 3600 Forbes Ave Suite 400A      _/   _/  _/      _/   _/   _/    :
: Pittsburgh, PA  15213             _/_/   _/        _/_/  _/_/_/  :
: 412-647-6852  fax:  412-647-5687                                 :
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Thomas Lumley
#
On Tue, 9 May 2000 richards at toutatis.pci.upmc.edu wrote:

            

      
      
      
    

        
You will find xyz.coords() in the development version of R, available from
CRAN. It won't be in a stable release until June sometime.

	-thomas

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Peter Dalgaard
    

    

      
      #
    

  


  

      
richards at toutatis.pci.upmc.edu writes:

            

      
      
      
    

        
...

            

      
      
      
    

        
Now that statement is quite a temptation to comment upon...

            

      
      
      
    

        
...especially since xyz.coords sits 5 lines down from 3D-Scatterplot
on Uwe's homepage!

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Uwe Ligges
    

    

      
      #
    

  


  

      
Peter Dalgaard BSA wrote:

            

      
      
      
    

        
... and also included in the *package* scatterplot3d_0.2.11.tar.gz (or
.zip for windows).

Regards,
Uwe Ligges
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Douglas Bates
    

    

      
      #
    

  


  

      
Ben Bolker <ben at zoo.ufl.edu> writes:

            

      
      
      
    

        
The generic profile is in the nls library, as are profile.nls and
plot.profile.nls.  Try 
 example(plot.profile.nls) 
to see the effect.
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Ben Bolker
    

    

      
      #
    

  


  

      
I've seen profile.nls; I was thinking about extending it to do general
likelihood surfaces, but it looks like all I need to is write a "profile"
method for generalized likelihoods (I don't know whether I should expend
the effort to follow the format of ms() as defined in S(-PLUS?) or not),
and "plot.profile" will work automatically.