[R-sig-ME] lme4a and Profile

Ben Bolker bbolker at gmail.com
Thu Feb 23 22:12:22 CET 2012


Rachel Cohen <stat.list at ...> writes:

> Hi, I would like to try using the profile() function etc.
> to look at confidence intervals on mixed model
> parameters (as detailed in Douglas Bate's online book).  
> Working through the book's example I get the
> following error message:

> Error in UseMethod("profile") : 
>   no applicable method for 'profile' applied to an object of class "mer"

>  I believe I have to install lme4a in order to use these 
> new functions.  I can't seem to find any mention of
> this new version of lme4 on R-forge.
>  Does anyone know where I can obtain lme4a and if it's ready/reliable
> for use?

(1)  If you have the appropriate tools for building packages installed
you should be able to just

install.packages("lme4a",repos="http://r-forge.r-project.org",type="source")

(2) Otherwise, I would normally tell you to
see http://lme4.r-forge.r-project.org , but ... the binary versions
of lme4a there are somewhat out of date, so they won't be automatically
installed in R 2.14.  You can try poking around at 
http://lme4.r-forge.r-project.org/repos/bin/ , downloading versions for 
slightly older versions of R and installing the binaries locally --
it *might* work.

(3) lme4a is relatively stable, lme4Eigen (the new, new, new
development version) is getting so -- it should run all the
profiling examples in the book, although Doug Bates is working
on updating a few small details in the book to make them work
with lme4Eigen.  You could try lme4Eigen ...

(4) if all else fails, ask again here for lme4a binaries to be
provided (I'd slightly rather spend my effort providing binaries
for lme4Eigen, but would be willing to make some new lme4a binaries
in the interim).




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