[R-sig-ME] Question about random effects variance in lme4

[Phillip Alday]

The same functionality in MixedModels.jl might be easier to understand,
especially with the math included in the docstring:

https://github.com/JuliaStats/MixedModels.jl/blob/628cbf31aac13f26dee4d2e2c50a0e459508d880/src/linearmixedmodel.jl#L269-L297

There is also an update in progress to allow it to work on more
complicated random effects:

https://github.com/JuliaStats/MixedModels.jl/pull/492/files

Phillip


On 4/5/21 4:29 pm, Chiara Masci wrote:
> Good morning,
> 
> I am a researcher at the Department of Mathematics of Politecnico di Milano, Italy.
> I am working in the field of parametric and non parametric mixed-effects models and I would like to ask a question about the computation of random effects in the R package lme4.
> 
> 
> If possible, I would like to ask a clarification about the random effects variance: I understand how the conditional variances of random effects are computed and then displayed by means of ranef and dotplot functions.  What I did not get is how the condVar values in ranef are related to the random effects variance that I can read from the summary of the lmer model (or, equivalently, from the function varCorr(lmer model)). I see that, if groups have the same number of observations within them, the values of condVar are all the same, so I guess they depend somehow on the number of observations. I am missing the relation between the group- specific condiational variances and the overall variance of random effects.
> 
> 
> Thank you very much!
> 
> Best wishes,
> Chiara Masci
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>