[R-sig-ME] Model fit with glmmadmb

Discussion:

Frederico Mestre

2018-07-18 14:23:36 UTC

Hello everybody,

I've been using
the
package glmmADMB and I wonder if I can get any information on the overall
fit of my models (
some sort of
pseudo R^2?).

I ran the function glmmadmb with negative binomial.

Cheers,

--

Frederico Mestre
ᐧ

[[alternative HTML version deleted]]

Mollie Brooks

2018-07-19 12:37:02 UTC

Permalink

Hi Frederico,

I don’t know about an R^2 with glmmadmb, but it is possible with glmmTMB for negative binomial GLMMs. I believe that glmmTMB has all the functionality of glmmadmb and more.

The r2 function is implemented in the sjstats package version 0.16.0.
https://cran.r-project.org/web/packages/sjstats/index.html

If you want to know some details such as how it doesn’t work with zero-inflated models yet, there is a discussion here
https://github.com/glmmTMB/glmmTMB/issues/169

cheers,
Mollie

Post by Frederico Mestre
Hello everybody,
I've been using
the
package glmmADMB and I wonder if I can get any information on the overall
fit of my models (
some sort of
pseudo R^2?).
I ran the function glmmadmb with negative binomial.
Cheers,
--

Frederico Mestre
ᐧ
[[alternative HTML version deleted]]
_______________________________________________
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

Paul Johnson

2018-07-19 12:37:50 UTC

Permalink

Hi,

The rsquared function in the piecewiseSEM package will do this for negative binomial GLMMs fitted with glmer.nb, but not (AFAIK) with glmmADMB. If you can’t refit your model with glmer.nb then you could DIY, following the methods in this paper (which are the same methods used by piecewiseSEM):

The coefficient of determination R2 and intra-class correlation coefficient from generalized linear mixed-effects models revisited and expanded
Nakagawa et al.
doi.org/10.1098/rsif.2017.0213

Appendix S6 section 2 shows a worked example for glmmadmb with negative binomial (family = “nbinom2”).

Best wishes,
Paul