[R-sig-ME] glmer not supported class for vuong()closeness test (non-nested models)

[Souheyla GHEBGHOUB]

Hi Mollie,

Thank you for your reply.
I want to test whether 3 predictors are significantly different from each
other when fitted to the same data but in different models by basically
comparing the models, like when we do the likelihood ratio test but that
one fits for nested models only.

Best,
Souheyla

On Fri, 7 Feb 2020 at 10:57, Mollie Brooks <mollieebrooks using gmail.com> wrote:

> If you want to use vuong() to test for zero inflation, then an alternative
> would be to use testZeroinflation() from the DHARMa package which supports
> glmer models. It performs a different test using simulated residuals.
>
> See documentation here
> https://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html
>
> Cheers,
> Mollie
>
>
> On 7Feb 2020, at 11:25, Souheyla GHEBGHOUB <souheyla.ghebghoub using gmail.com>
> wrote:
>
> Hi,
>
> I need to compare non-nested models fitted to the same DV using vuong()
> closeness test in pscl package. It appears that the test only supports glm
> class but not mixed models glmer.  I would like to retain glmer because my
> random effects explain much of the variation but vuong does not support it!
>
> Should I simply ignore the random effects at this point since they will be
> equally removed from both compared models in this sig test, is there a way
> to force the test to support this glmer class?
>
> Thanks a lot for your time reading/answering this,
>
> Souheyla GHEBGHOUB
> 4th year PhD student
> Education department - Research Centre for Social Science
> University of York
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>
>

	[[alternative HTML version deleted]]