[R-sig-ME] Variance explained by random factor

Gregor Gorjanc gregor.gorjanc at bfro.uni-lj.si
Thu Aug 14 09:29:48 CEST 2008

Renwick, A. R. <a.renwick at ...> writes:
> However,the model output that I get states that the variance explained by the
random factor is 0:
> Generalized linear mixed model fit by the Laplace approximation
> Formula: RoundedOverlap ~ sess + breedfem + sess:breedfem + (1 | Site)
>    Data: Male
>    AIC   BIC logLik deviance
>  109.9 127.2 -45.93    91.86
> Random effects:
>  Groups Name        Variance Std.Dev.
>  Site   (Intercept)  0        0
> Number of obs: 51, groups: Site, 14

I smell a very flat likelihood here. You have 51 records, and you would
like to fit some "fixed" effects (you do not show the whole output so we
can not evaluate how many levels are there for sess and breedfem) and
a "random" effect of Site. I guess there is very little information in 
the data for the stimation of variance between the sites.


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