[R-sig-ME] Likelihood drops on adding random effect

David Duffy David.Duffy at qimr.edu.au
Thu May 17 02:22:30 CEST 2012


On Thu, 17 May 2012, Murray Jorgensen wrote:

> it is counterintuitive to me that adding a random effect to a GLMM should
> lead to a drop in the likelihood of the fitted model

> f0128bit: fincr ~ icfac + (1 | gap) + (1 | box) + (1 | gap:box) + (1 |
> f0128bit:     tree)
>         Df    AIC    BIC  logLik Chisq Chi Df Pr(>Chisq)
> f0128bi   7 615.73 655.48 -300.87
> f0128bit  8 687.01 732.43 -335.51     0      1          1

It's not counterintuitive, it's wrong  ;)  So either N's are different or 
program got stuck.  Try fiddling with starting values.



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