[R-sig-ME] Reply: Inquiry about GLMM model

[Caroline Zanchi]

But you do have several treatments where some individuals die ?

If no individual died in one level of your variable, then simply say "no 
individuals died in treatment ...", and interpret the rest of your data.

Best,

Caroline

On 6/30/23 16:32, Yu qing Fan wrote:
> Hi,
>
> Yes, that is my case! All my individuals survived at last. Do you have 
> any recommendations for such scenario? Maybe a simpler modelling could 
> help?
>
> Thanks!
> Yuqing Fan
>
>
> ------------------------------------------------------------------------
> Caroline Zanchi <caroz using zedat.fu-berlin.de>
> Yu qing Fan <yqfan using uwaterloo.ca>
> *R*e: [R-sig-ME] Inquiry about GLMM model
> Hello !
>
> This can happen if there is a complete separation of the outcome for the
> response variable per explanatory variable level. For example, if you
> check proportion of surviving individuals and in one treatment all of
> them die, or on the contrary all of them survive.
>
> Is it your case by any chance ?
>
> Caroline
>
> On 6/27/23 17:16, Yu qing Fan wrote:
> > Hello,
> >
> > Hope you all had a nice day. My name is Yuqing Fan and I am a 
> Bachelor student. I am doing a longitudinal analysis using GLMM model, 
> and now I encounter some diffculties.
> >
> > As you might see in the codes, when generating the first 
> model(fit1), R would say "function evaluation limit reached without 
> convergence". I reduced the number of covariates like fit2 and fit3, 
> and that's how I found the covaraite "position" led to error. I then 
> cut levels of "position" from 5 to 3, yet the error still occurs. 
> Could anyone please give me some advice about how can I fix problems 
> here? Thank you for your time!
> >
> > Yours sincerely,
> > Yuqing Fan
> > _______________________________________________
> > R-sig-mixed-models using r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
	[[alternative HTML version deleted]]