[R-sig-ME] Error using glmmADMB

Ben Bolker bbolker at gmail.com
Wed Aug 1 01:40:55 CEST 2012

Belén Fresnillo <belen.fresnillo at ...> writes:

> Dear all list members,
> I´m trying to use glmmADMB package to analize my count data in interactions
> between adult and subadult
> lizards. I have repeated measures zero inflated count data from videotaped
> interactions of different
> durations, so I´m trying to use glmmADMB with an offset variable (recording
> time), but it doesn't seem
> like working. I keep finding this error message with any of the models
> including the offset: 
> >
> model<-glmmadmb(interactions~treatment*sex.sub*sex.adult+
>   svl+offset(recording.time)+(1|id.adult),
> zeroInflation=TRUE, family="poisson")

  It's extremely likely that you meant offset(log(recording.time)) ; offsets are
on the scale of the linear predictor (log in this case).  I don't know if that
will help (of
course as you point out it doesn't help with NB1).

  What happens if you try this (1) with glm(), with neither zero inflation nor
random effects;
(2) with pscl::zeroinfl, without random effects; (3) with glmmADMB, without the
If these all give reasonable answers but the full model still fails, see some of
the debugging tips
in ?admbControl ...

  The key to giving enough information is to give a *reproducible* example if at
all possible
(see http://tinyurl.com/reproducible-000 ...) -- the results of sessionInfo()
are helpful too
so we can see what version of R/packages you're using.

  Ben Bolker

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