[R-sig-ME] [FORGED] Re: Question about gls

Berwin A Turlach berwin.turlach at gmail.com
Sat Mar 24 07:36:33 CET 2018

G'day Rolf,

On Sat, 24 Mar 2018 10:52:34 +1300
Rolf Turner <r.turner at auckland.ac.nz> wrote:

> On 24/03/18 08:58, Andrzej Galecki wrote:
> > Hi Louise,
> > 
> > One possible way ...
> Not exactly obvious, but, is it?  How on earth would one ever figure 
> this out, except by appealing to r-sig-mixed-models?

Well, in the good old days, over on r-help, whenever people complained
that it was not obvious how to do something using the MASS package they
were gently but firmly reminded by a certain person that the MASS
package is support software for a certain book and that they should
read the fine book (RTFB). :)

I believe Doug Bates is far to polite to point out the the nlme package
is also support software for a certain book and that you need to RTFB
for the gory details.....

> Looking at str(fit.gls) one sees that there is a (list) component 
> "modelStruct" and that this component in turn has a component 
> "corStruct".  If one is an optimist, one might try
>      fit.gls$modelStruct$corStruct
> which (mirabile dictu!) gives:
> Again if one is an optimist, one might try stripping away all the 
> unwanted baggage by doing
>      as.vector(fit.gls$modelStruct$corStruct)

> Still it's all pretty obscure.  And typical of the obfuscation that 
> results from using S4 classes and methods.

I see that Doug has corrected you already, but the fact that you are
typing `$` all the time instead of `@` should have been a dead give
away that this is S3, not S4.  Of course, the length of time that the
nlme package is around is another give away....



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