Skip to content

error message in glmmADMB

2 messages · Marcela Rodriguero, Ben Bolker

#
Dear list members,

I am trying to fit a glmmADMB model on my data, but altough I tried to fix
it after reading many tutorials and forums, I couldn't do it.
My response variable is the clutch size of females treated with an
antibiotic or untreated (control females), and I took data over many days.
Because lots of females put 0 eggs, I chose negative binomial over poisson
family.
The model looks like this:

model A= glmmadmb(count~treatment+(time|female),family="nbinom",data=table1)

This is the error message:
Parameters were estimated, but not standard errors were not: the most likely
problem is that the curvature at MLE was zero or negative
Error in glmmadmb(count ~ treatment + (time | female), family =
"nbinom",data=table1) : 
  The function maximizer failed (couldn't find STD file) Troubleshooting
steps include (1) run with 'save.dir' set and inspect output files; (2)
change run parameters: see '?admbControl'
In addition: Warning message:
running command 'C:\Windows\system32\cmd.exe /c
"C:/Users/Marce/Documents/R/win-library/3.0/glmmADMB/bin/windows64/glmmadmb.exe"
-maxfn 500 -maxph 5 -noinit -shess' had status 1 

"Count" is the clutch size, my response variable, 'treatment' is a fixed
effect, because I want to distinguish between clutch size of treated vs.
untreated females, and "time" and "female" are random effects because of the
pseudorreplicate (i.e. random effects as nuisance).

Any idea of what is wrong? Thanks in advance!
Marcela
1 day later
#
Marcela Rodriguero <rodriguero at ...> writes:
OK.  (You might want to try zero-inflated variants as well.)
Hard to say without seeing your data.  How big is your data set
(how many females, how many per treatment, approximately how many
measurements per female?)  Did you try a simpler model such
as ~treatment+(1|female) ?  Did you try out the trouble-shooting
suggestions listed in ?admbControl?  (Other standard troubleshooting
suggestions such as centering and scaling your data may be less
relevant since your design looks pretty simple.  However, if the
minimum value of your time variable is far from zero, you might want
to subtract the minimum or center the time variable)

  Ben Bolker