[R-sig-ME] Is there a final-ish format for lme4 objects now?

Martin Maechler maechler at stat.math.ethz.ch
Tue Apr 10 16:22:06 CEST 2012


>>>>> Paul Johnson <pauljohn32 at gmail.com>
>>>>>     on Mon, 9 Apr 2012 16:48:10 -0500 writes:

    > Greetings: Long ago, I wrote a function "outreg" to make
    > LaTeX tables for regressions.  Now I've put it in a
    > package called rockchalk.  Because I put it in a package
    > on CRAN, people I don't know write to me saying "it works
    > for lm and glm, but what about lme4".

    > In 2008 or so, I had tried to incorporate lme4 output, but
    > I ran into the trouble. Version changes of lme4 regularly
    > broke my output code.  I asked about it here, and learned
    > I was "jumping the gun" because lme4 was still adjusting.

    > Has lme4 output reached a point where I can treat the
    > output from lme4 functions as "more or less" stable?  If I
    > make the table maker function work with lme4 objects (the
    > stable CRAN version), is that function likely to also work
    > with the development version of lme4? 

Yes, if you do *not* access slots of the resulting object
directly,
but rather use methods such as
    residuals(), fitted(), coefficients(), varCorr()
*or* then, the 
    getME()  function which we've been providing for  CRAN-lme4
for a while now, to make the transition between CRAN-lme4
(will become "lme4.0" and basically unmaintained) and  R-forge-lme4
which will become the new CRAN-lme4
as soon as we have removed the few remaining show-stopping
glitches.

See this posting:
  https://stat.ethz.ch/pipermail/r-sig-mixed-models/2011q4/006786.html

and the examples in  ?getME  in CRAN-lme4.

    > I know the class of
    > the objects changed from mer to merMod.  Is that a signal
    > that I am again too early to try to do this?

no, because we tried to ensure that useRs and package writers
can build on top of old- or new-lme4, *not* accessing the
objects directly.

Martin

--
Martin <Maechler at stat.math.ethz.ch>  http://stat.ethz.ch/people/maechler
Seminar für Statistik, ETH Zürich  HG G 16      Rämistrasse 101
CH-8092 Zurich, SWITZERLAND
phone: +41-44-632-3408       fax: ...-1228      <><

    > pj
    > -- 
    > Paul E. Johnson Professor, Political Science  
    >  Assoc. Director 1541 Lilac Lane, Room 504     Center for
    > Research Methods University of Kansas              
    > University of Kansas http://pj.freefaculty.org          
    >  http://quant.ku.edu



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