s-news
[Top] [All Lists]

Degrees of freedom problem with nested glm

To: s-news@lists.biostat.wustl.edu
Subject: Degrees of freedom problem with nested glm
From: vspeters <vspeters@ualberta.ca>
Date: Sun, 23 Jun 2002 17:57:17 -0600
Hello,
     I appear to be having a problem with my degrees of freedom and parameters 
in the following model, however, I do not understand why.

Model:
glm(density ~ treatment/stand + pgstems, family = quasi(link = "log", var = 
"mu"), na.action = na.omit, data=chapter6, maxit = 20)

Message:
Error in glm.fitter(x = X, y = Y, w = ..: 109 parameters
 but only 105 observation

Data Breakdown:
There are 2 treatments
          56 stands - essentially a replicate(approximately 28 per treatment)
                    - 49 have 2 observations, the rest have 1
          105 observations in total.

Why would this model have more parameters then observations?  
Is there a different way of running the glm so multiple observations are still 
nested within the appropriate stand?

All responses are appreciated.

Vern Peters


<Prev in Thread] Current Thread [Next in Thread>
  • Degrees of freedom problem with nested glm, vspeters <=