[R-sig-ME] models with no fixed effects
ajmackey at gmail.com
Thu Sep 11 22:17:57 CEST 2008
On Thu, Sep 11, 2008 at 4:09 PM, Peter Dixon <peter.dixon at ualberta.ca> wrote:
> I think it would make sense to consider a model in which E[Y] = 0 when
> the data are (either explicitly or implicitly) difference scores. (In
> fact, I tried to fit such a model with lmer a few months ago and ran
> into exactly this problem.)
I had the same initial thoughts, but for me, the intercept tells me
how much of the difference remains unexplained by the covariates.
Fixing that at 0 is the same as infinitely strong prior belief that
there are no other possible explanations for the observed differences.
More information about the R-sig-mixed-models