Skip to content

extracting residuals in lmer glmm

3 messages · Martin Henry H. Stevens, Douglas Bates

Original
Martin Henry H. Stevens
# 
Hi folks,
I seem unable to figure out how to extract residuals from a Poisson  
lmer model.

And, I am not sure what to do with the working residuals.

Any references or ideas would be appreciated.

# Example:
y <- rpois(100, lambda=10)
xz <- expand.grid(x=gl(2,5), z=gl(10,1))
m1 <- lmer(y ~ x + (1|z), xz, family="poisson")

resid(m1)
Error: 'resid' is not implemented yet

  residuals(m1)
Error: 'residuals' is not implemented yet

  summary(m1 @ wrkres)
Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
1.38    1.98    2.20    2.22    2.50    3.13


Hank Stevens
Martin Henry H. Stevens
# 
In answer (sort of) to my own question, resid() now works on both mer  
and summary.mer objects.
All of these results (resid() working and not working) have occurred  
while using yesterday's CRAN version of lme4 {R package version  
0.999375-24}.

I have no idea what would have caused the problem yesterday and this  
morning, but not now. Oh well.

Hank


y <- rpois(100, lambda=10)
xz <- expand.grid(x=gl(2,5), z=gl(10,1))
m1 <- lmer(y ~ x + (1|z), xz, family="poisson")
m1s <- summary(m1)
resid(m1)
   [1] -0.04336 -0.98556  0.89885  0.27071 -0.67149 -0.14260
...
resid(m1s)
   [1] -0.04336 -0.98556  0.89885  0.27071 -0.67149 -0.14260
...
On Aug 14, 2008, at 9:32 AM, Martin Henry H. Stevens wrote:

            

      
      
      
    

        
Dr. Hank Stevens, Associate Professor
338 Pearson Hall
Botany Department
Miami University
Oxford, OH 45056

Office: (513) 529-4206
Lab: (513) 529-4262
FAX: (513) 529-4243
http://www.cas.muohio.edu/~stevenmh/
http://www.cas.muohio.edu/ecology
http://www.muohio.edu/botany/

"If the stars should appear one night in a thousand years, how would men
believe and adore." -Ralph Waldo Emerson, writer and philosopher  
(1803-1882)

  
  


  


      

    

    
1 day later

    

      
      

        


  

        

      Douglas Bates
    

    

      
      #
    

  


  

      
On Thu, Aug 14, 2008 at 9:47 AM, Hank Stevens <HStevens at muohio.edu> wrote:

            

      
      
      
    

        
I was just going to write that you must be using a very old version of
the lme4 package.  The wrkres slot was removed in the transition from
the 0.99875 to 0.999375 series.

What is returned by the resid generic applied to a generalized linear
mixed model is

(m1 at y - m1 at mu)/sqrt(m1 at var)

That is one of the many possible definitions of "residuals" for a
generalized linear model.  There is a name for those residuals but I
can't remember right now what they are callled - perhaps after I have
recovered from the trans-Atlantic flight returning from the useR!2008
conference, which was wonderful.  Thanks again to the organizers and
to the more than 400 attendees who made it such a success.  We hope to
see many of you again in Rennes, France for useR!2009 next July.