Message-ID: <20120925215511.GA5805@samir>
Date: 2012-09-25T21:55:11Z
From: Hans Ekbrand
Subject: Interpreting the output of summary() of a glmer-object
In-Reply-To: <loom.20120925T234007-68@post.gmane.org>
On Tue, Sep 25, 2012 at 09:43:48PM +0000, Ben Bolker wrote:
> Hans Ekbrand <hans at ...> writes:
> > So, if I understand you - which I think I do - then 0.63685 is
> > simply the standard deviation of ranef(my.fit.1)?
> >
> > When I try to compute that manually I get a numerically
> > close figure, but not quite the same:
> >
> > sd(unlist(ranef(my.fit.1)))
> > [1] 0.6423346
> >
> > I am on the right track?
> >
>
> I think this has been discussed previously on the list (although
> I can't point you to a precise spot), but:
>
> http://permalink.gmane.org/gmane.comp.lang.r.lme4.devel/8580
>
> says
>
> "Generally you will find that the variance of the values returned by
> ranef is less than the estimated variance because of shrinkage."
>
> Sorry I can't go into more detail right now.
No need to, but thanks for the pointer. I take that it's not *simply*
the standard deviation, but it is the standard deviation :-)