[R] variance using lmer

[Spencer Graves]

	  I'm convinced that I do NOT understand your question, but the most 
probable answer given my prior is "no".  To get a more specific answer, 
please provide more detail about the problem you are trying to solve, 
preferably including a simple, self-contained example, e.g., the 
examples in the "lmer" and "lmer-class" help pages.  If the CarCorr and 
vcov functions do not answer your questions, please explain why not, 
e.g. using for illustration a 'predict' example applied to lme {nlme} or 
glm.

	  I realize this does not answer your question, but I hope it gets you 
closer.

	  Buena Suerte,
	  spencer graves

Per Theilgaard wrote:
> Dear R help
> I have a question on the variance of the binomial probit model.
> 
> I have fitted the following model :
> 
>> lmer1<-lmer(mp ~ l + op + l*op+ us_lev + bw_lev +(1|tatu) ,  
> +                family = binomial(link="probit"), 
> +                method = 'Laplace',
> +                data = matings,
> +                msVerbose= True)
>> summary(lmer1)
> Generalized linear mixed model fit using Laplace 
> Formula: mp ~ l + op + l * op + us_lev + bw_lev + (1 | tatu) 
>    Data: matings 
>  Family: binomial(probit link)
>       AIC      BIC    logLik deviance
>  1455.121 1556.387 -706.5605 1413.121
> Random effects:
>      Groups        Name    Variance    Std.Dev. 
>        tatu (Intercept)     0.42905     0.65502 
> # of obs: 918, groups: tatu, 130
> 
> Estimated scale (compare to 1)  0.8521378 
> 
> I suppose that the dispersion parameter has been restricted to be 1.
> Then the variance would be 1 + the variance of the random effect: 1+ 0.42905 ???
> 
> Thany you very much in advance
> Per
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html