[Rd] odd evaluation within correlation argument of glmmPQL

Ben Bolker bolker at ufl.edu
Tue Oct 27 20:05:14 CET 2009




Ben Bolker wrote:
> 
> 
>  [snip]
> 
>   It appears that glmmPQL looks in the global workspace, not
> within the data frame specified by the "data" argument, for
> the variables specified in the "form" argument of spatial
> correlation structures provided to the "correlation" argument.
> 
> [snip example and patches]
> 
> 

Replying to my own e-mail to bump this (it's been a week).

 The basic issue is that glmmPQL throws out data that are
not involved in the fixed or random model terms, or in the offset
term.  It doesn't save variables that are only found in the
correlation formula argument.  It's hard to imagine how
this behavior is anything other than a bug ... and the patch fixes it
in a straightforward way.

  cheers
   Ben Bolker

-- 
View this message in context: http://www.nabble.com/odd-evaluation-within-correlation-argument-of-glmmPQL-tp25986187p26083255.html
Sent from the R devel mailing list archive at Nabble.com.



More information about the R-devel mailing list