[R-sig-ME] Questions concerning glmmADMB and hurdle model

Thierry Onkelinx thierry.onkelinx at inbo.be
Tue Apr 5 09:22:34 CEST 2016

Dear Amirouche,

A common misconception is that a lot of zero's in the the response
automatically implies zero-inflation. Poisson or negative binomial
distributions with low expected values yield a lot of zero's, even without

So my advice would be to fit a Poisson or negative binomial model and test
if there is any zero-inflation.

Best regards,

ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey

2016-03-30 11:46 GMT+02:00 Amirouche Sadoun <
amirouche.sadoun op cerco.ups-tlse.fr>:

> Dear Sir,
> I'm currently a PhD student in cognitive neurosciences at the CerCo
> laboratory, France.
> I'm dealing with a problematic trying to analyze my data with the glmmADMB
> function in R.
> I thank you in advance for paying attention to my present problem
> regarding the use of a Hurdle model with glmmADMB, because I didn't yet
> found a solution, especially as I am new in using R.
> I'm trying to analyze behavioral data containing a given number of trials
> called ab. (Experimental design with repeated measures.  Fixed effects:
> GROUP and DELAY; random effect: the id of the subjects). (Please find the
> data attached).
> The response variable is the number of AB responses related to the Total
> number of trials.
> In this data, many subjects had 0 values . This requires the use of a Zero
> inflated or Hurdle model with glmmADMB function (), as it was recommended
> to me.
> I tried than to do the analysis with this function, even if it seems not
> clear to me, but I get error messages, in addition to that I don't know how
> to deal with both models after (with post hoc analysis).
> Please, find the following lines of the R code and the error messages:
> datab $ ID <-factor ($ datab ID)
> mod1 <-glmmadmb (cbind (AB, Total-AB) ~ * DELAY GROUP + (1 | ID), data =
> subset (datab, AB> 0), family = "truncnbinom1")
> datab $ nz <- as.numeric (datab $ AB> 0)
> mod2 <-glmmadmb (nz ~ * DELAY GROUP + (1 | ID), data = datab, family =
> "binomial")
> The error message: [1] NOTICE: Warning in eval (expr, envir, pen):
> sd.est not defined for this family
> [2] ERROR:
> The maximizer function failed (could not find parameter file)
> Troubleshooting steps include (1) run with
> 'Save.dir' set and inspect output files; (2) changes run parameters
> I hope wholeheartedly find help
> Please accept the assurance of my distinguished regards.
> CerCo, UMR 5549
> Pavillon Baudot
> Toulouse 31052 FRANCE
> _______________________________________________
> R-sig-mixed-models op r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

	[[alternative HTML version deleted]]

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