[R-sig-ME] mcmcsamp(lme4) & mcsamp(arm) & HPDinterval(coda): 2 questions

Marco Demont demont at access.uzh.ch
Thu Apr 24 15:42:17 CEST 2008

Dear lme4 users,

I want to generate a sample from the posterior distribution of the 
parameters of my fitted mixed model using the mcmcsamp function (lme4 
Package). Additionally, I would also like to create 95% highest posterior 
density (HPD) intervals for the parameters in the MCMC sample (maybe with 
the HPDinterval function from the coda Package?).

For that purpose I got two questions:

Question 1:
How can I adjust or change the number of chains, the thinning interval and 
the burn-in for the mcmcsamp function (lme4 Package)?

Question 2:
I did this (e.g. change number of chains, thinning interval, etc) in the 
mcsamp function (arm Package) (mcsamp relies on mcmcsamp). mcsamp creates an 
object of class “bugs”. How can I get 95% highest posterior density (HPD) 
intervals for this “bugs” object?

To summarize my problem: in my first approach (Question 1) I’m able to 
create HPD intervals, but NOT to change number of chains, thinning interval 
etc; in my second approach (Question 2) I’m able to change number of chains, 
thinning interval etc, but NOT to create HPD intervals.

Would be nice if somebody could help!


Marco Demont
Zoological Museum
University of Zuerich
Winterthurerstrasse 190
CH-8057 Zuerich

Tel.: +41 44 635 47 79
demont at access.uzh.ch
demont at gmx.ch

More information about the R-sig-mixed-models mailing list