R help - Jan 2015 - glmmADMB model fails when fixed effect interaction term is removed

Hello,

I am looking at a data set predicting number of plants (FERO) based on fixed
effects sowing method (Method) and sowing rate (Rate). I also have a slope
random effect for the site and intercept random effects for the nested
variables Block and Strip.ID (to account for experimental design). I have
tried running several different models using the glmmadmb function. The
dependent variable, FERO, is a discrete count. All of my fixed and random
effects are categorical and have the following levels:

Method: 3
Rate: 5
Site: 3
Block: 18
Strip.ID: 275

The following model runs without a problem:

fit_nbin3 <-glmmadmb(FERO ~ Method * Rate + (FERO|Site) +
(1|Block/Strip.ID),
data = data1, family = "nbinom")

However, when I try to run the same model without the interaction term:

fit_nbin4 <-glmmadmb(FERO ~ Method + Rate + (FERO|Site) +
(1|Block/Strip.ID),
+ data = data1, family = "nbinom")

I get the following 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(FERO ~ Method + Rate + (FERO | Site) + (1 |
Block/Strip.ID),  : 
  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/Cara/Documents/R/win-library/3.1/glmmADMB/bin/windows64/glmmadmb.exe"
-maxfn 500 -maxph 5 -noinit -shess' had status 1

I am confused as why removing the interaction term, which should simplify
the model is causing it to fail to run correctly. Any ideas what's going on
here?

Here is a sample of my first few lines of data:

Site         Method         Rate  Block  Strip  FERO
Glacial      Broadcast      A      A        H     69	
Glacial      Broadcast      A      A        H     1
Glacial      Broadcast      A      A        H     41

One other quick thing:
I can't seem to the the aictab function to work on the properly run glmmadmb
models. I get a message saying: 	
Error in formatCands(cand.set) : 
Functions do not support mixture of model classes

Thanks in advance for your help!

Cara Applestein
Prairie Science Technician
Center for Natural Lands Management
Olympia, Washington



--
View this message in context:
http://r.789695.n4.nabble.com/glmmADMB-model-fails-when-fixed-effect-interaction-term-is-removed-tp4701411.html
Sent from the R help mailing list archive at Nabble.com.

Cara,

Perhaps I a bit ignorant about mixed models, but I don?t understand why you have
the dependent variable included as a random effect.

The variable names you are using suggest an analysis of combined trials, so (1 |
Site/Block/Strip.ID) seems more appropriate. Is this a factorial or a
split/strip-block design?

Peter
> On Jan 5, 2015, at 3:14 PM, cara715 <capplestein at cnlm.org> wrote:
> 
> Hello,
> 
> I am looking at a data set predicting number of plants (FERO) based on
fixed
> effects sowing method (Method) and sowing rate (Rate). I also have a slope
> random effect for the site and intercept random effects for the nested
> variables Block and Strip.ID (to account for experimental design). I have
> tried running several different models using the glmmadmb function. The
> dependent variable, FERO, is a discrete count. All of my fixed and random
> effects are categorical and have the following levels:
> 
> Method: 3
> Rate: 5
> Site: 3
> Block: 18
> Strip.ID: 275
> 
> The following model runs without a problem:
> 
> fit_nbin3 <-glmmadmb(FERO ~ Method * Rate + (FERO|Site) +
> (1|Block/Strip.ID),
> data = data1, family = "nbinom")
> 
> However, when I try to run the same model without the interaction term:
> 
> fit_nbin4 <-glmmadmb(FERO ~ Method + Rate + (FERO|Site) +
> (1|Block/Strip.ID),
> + data = data1, family = "nbinom")
> 
> I get the following 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(FERO ~ Method + Rate + (FERO | Site) + (1 |
> Block/Strip.ID),  : 
>  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/Cara/Documents/R/win-library/3.1/glmmADMB/bin/windows64/glmmadmb.exe"
> -maxfn 500 -maxph 5 -noinit -shess' had status 1
> 
> I am confused as why removing the interaction term, which should simplify
> the model is causing it to fail to run correctly. Any ideas what's
going on
> here?
> 
> Here is a sample of my first few lines of data:
> 
> Site         Method         Rate  Block  Strip  FERO
> Glacial      Broadcast      A      A        H     69	
> Glacial      Broadcast      A      A        H     1
> Glacial      Broadcast      A      A        H     41
> 
> One other quick thing:
> I can't seem to the the aictab function to work on the properly run
glmmadmb
> models. I get a message saying: 	
> Error in formatCands(cand.set) : 
> Functions do not support mixture of model classes
> 
> Thanks in advance for your help!
> 
> Cara Applestein
> Prairie Science Technician
> Center for Natural Lands Management
> Olympia, Washington
> 
> 
> 
> --
> View this message in context:
http://r.789695.n4.nabble.com/glmmADMB-model-fails-when-fixed-effect-interaction-term-is-removed-tp4701411.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.