[R-sig-ME] Mcmcsamp and Confidence Intervals

[Jörg Albrecht]

Hi,

you can use the function "pvals.fnc" from the package languageR to 
obtain the posterior distributions of the parameters of interest.

Best wishes,

Jörg Albrecht

Am 07/07/2013 19:16, schrieb tommy gaillard:
> Dear all,
>
>   I am aiming to assess the between-individual variability in random slope
> and random intercept in different context. Here is my main model:
>   fit<-lmer(logitprop_vig~logdensity+logbiomasstrue+loggpsize+position+(1|Idtag),na.action=na.omit,
> data=tabimpala)
>
> I would like to obtain for each of the explanatory variables the Vi
> (variability in random intercept) and Vs (variability in random slope) with
> their CI 95%.
>
>   Firstly I have read in many post that the mcmcsamp function caused some
> problems..Is it still the case or can I use it without getting any biaises?
>
>   Secondly, I did not manage to obtain the IC95% of the parameter estimates
> in R.
> Here is what I did:
> fit<-mcmcsamp(fit,n=100)
> profile(fit)
> HPDinterval(fit)
>   With both functions above I got: Error in UseMethod :
> method for 'profile/HPDinterval' not applicable for an object of class
> "merMCMC"
>
> This is really important to me. I hope that you help me out!
>
>   Many thanks!
>
>


-- 
M.Sc. Jörg Albrecht

Department of Ecology - Conservation Ecology
Faculty of Biology
Philipps-Universität Marburg
Karl-von-Frisch-Str. 8
35043 Marburg
Germany

mail: joerg.albrecht at staff.uni-marburg.de
phone: +49 6421 28 25385
fax: +49 6421 28 23387