[R-sig-ME] varIdent in glmmTMB?

[Ben Bolker]

   The glmmTMB analogue of varIdent(form = ~ Temperature) would be

dispformula = ~ Temperature

   or (if you want the variance model to be parameterized as a 
log-dispersion for each temperature rather than a log-dispersion for the 
first temperature level plus differences in log-dispersion)

   dispformula = ~ 0 + Temperature

   cheers
    Ben Bolker

On 2025-01-14 7:14 a.m., Davide Nespoli wrote:
> Dear development team,
> 
> this is Davide Nespoli, researcher at the University of Florence (Italy).
> After building a zero-inflated negative binomial GLMM of the form:
> 
> m<-glmmTMB(Time~Temperature*Sex+Position+(1|Individual),family=nbinom1(),
> ziformula=~1, data=Perm)
> 
> I have a good model, but with a bit of heteroschedasticity due to the
> different levels of Temperature (which is a 3-level factor).
> I am wondering if it is possible to integrate a varIdent covariance
> structure in the model.
> I have already tried
> "+(1|Temperature), dispformula=~0,"
> but it doesn't work (neither for ZINB nor for ordered beta on proportional
> data).
> Is there a way to integrate it in such a model in glmmTMB?
> 
> Thank you very much.
> Best regards,
> 
> 
> 
> --
> Davide Nespoli, PhD
> Department of Biology
> University of Florence
> via Madonna del Piano 6, I-50019
> Sesto Fiorentino (FI), Italy
> email: davide.nespoli using unifi.it
> Tel. +390554574722
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models