[R-sig-ME] missing data in lme, lmer, PROC MIXED

[Douglas Bates]

On Fri, Jul 25, 2008 at 4:28 PM, M Henry H Stevens <HStevens at muohio.edu> wrote:
> Hi folks,
> I have colleagues who comfortably state that "missing data" are ok in
> "mixed models" - because "the program (PROC MIXED) handles missing data
> -- I have a hard time imagining what it does.

> To those of you who use both R and SAS, I was wondering if you might
> share insight into what these do.

Perhaps they are referring to a formulation of models for repeated
measures data where you view the responses as in the form of a matrix
for subjects and occasions.  Similarly, in item response models the
data are often represented as a matrix.  In that case one can have
missing cells in the matrix but those observations are not present in
any way when you switch from the "wide" format to the "long" format.
(See ?reshape for more information on wide and long formats for
repeated measures data.)

> As far as I know, for lme:
> 'na.action="na.omit" ' or na.exclude, removes the rows with any missing
> data.

Yes.  And that's the only sensible way of handling missing cells in
the matrix if you want to view the data in the wide format.  However
some old computational methods always assumed that you had a full
matrix layout and had to fudge the case of missing cells in that
layout.