[R-sig-ME] Zero inflated negative binomial mixed models - glmmADMB help

GERMAIN MARION marion.germain at univ-lyon1.fr
Fri Mar 11 10:29:32 CET 2016

Dear all,
I am trying to fit zero inflated negative binomial models (using the package glmmADMB) but with no success.
My response variable is the lifetime reproductive success of a bird species.
The explanatory variable are sex (M vs. F), natal dispersal status (D vs. P), age (Y vs. O), brood size manipulation status (three levels) and the one order interaction between natal dispersal status and brood size manipulation as fixed effects. I also include a random effect which is the first year of breeding for a bird.

I have several errors messages such as:
Parameters were estimated, but standard errors were not: the most likely problem is that the curvature at MLE was zero or negative
Erreur dans glmmadmb(LRS ~ sex + nataldisp + agecat + manipcat.control +  :
  The function maximizer failed (couldn't find parameter file) Troubleshooting steps include (1) run with 'save.dir' set and inspect output files; (2) change run parameters: see '?admbControl';(3) re-run with debug=TRUE for more information on failure mode
Furthermore : Warning message :
l'exécution de la commande './glmmadmb -maxfn 500 -maxph 5' renvoie un statut 1

I have been trying several solutions you already suggested such as:
- admb.opts=admbControl(shess=FALSE,noinit=FALSE)
- admb.opts=admbControl(shess=FALSE,noinit=FALSE, impSamp=200,maxfn=1000,imaxfn=500,maxph=5)

I also tried to downgrading the package to an earlier version (0.7) but I did not succeed. I followed your instruction (found here http://lists.admb-project.org/pipermail/users/2012-January/001667.html) but it doesn’t work.

I therefore have several questions. First, how could I know if I need to use a zero inflated model (until now, I used negative binomial mixed models). Second, do you have an idea of the reason of this error message ? I tried a zero inflated with a poisson distribution and I do not have any error message. I was wondering if the model could be overparametrized (n = 827). Finally, what could I do to resolve these problems ?
Thank you for helping,

	[[alternative HTML version deleted]]

More information about the R-sig-mixed-models mailing list