Douglas Bates <bates <at> stat.wisc.edu> writes: ..... > I encourage users of lmer who wish to determine the precision of the > estimates of the variance components to create a Markov chain Monte > Carlo sample of the parameters and evaluate the HPDintervals. > > > sm1 <- mcmcsamp(fm1, 50000) > > library(coda) > Warning message: > use of NULL environment is deprecated > > HPDinterval(sm1) > lower upper > (Intercept) 236.6518363 266.5465536 > Days 7.0136243 13.8947993 > log(sigma^2) 6.2550082 6.7295329 > log(Sbjc.(In)) 5.4928205 7.5751372 > log(Sbjc.Days) 2.8197523 4.6337518 > atanh(Sbj.(I).Dys) -0.6988632 0.9836688 > deviance 1752.5158501 1766.6461469 > attr(,"Probability") > [1] 0.95 >
And DB wrote in http://finzi.psych.upenn.edu/R/Rhelp02a/archive/76742.html "Evaluating entire terms is more difficult but you can always calculate the F ratio and put a lower bound on the denominator degrees of freedom." As a mcmc-challenged subject, it would be nice if someone could provide an example for this; or how to get CI estimates for an arbitrary contrast with mcmcsamp. Dieter ______________________________________________ R-help@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