[R-sig-ME] Does MCMCglmm allow offset?

Gustaf Granath Gustaf.Granath at ebc.uu.se
Tue Dec 21 19:25:18 CET 2010


Hi,
The offset term should have mean = 1, I guess. A work around might be to 
fix the parameter to one. Or?
The results corresponds to what lmer() gives.

prior2 = list(B= list (mu = matrix(c(0,1,0,0,0),5),V = diag(5)*1e+6))
diag(prior2$B$V)[2]<-(1e-6)      #change the prior variance to something 
small
mcmc.off2 <- MCMCglmm(rapi ~ log(off30)+gender*time, data =
                 rapi.df,family = "poisson", verbose = TRUE,
         random = ~ us(1 + time):id,prior=prior2)
summary(mcmc.off2$Sol)

Please correct me if Im wrong.

Gustaf




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