[R-sig-ME] Calculating F values for lme function

Ben Bolker bbo|ker @end|ng |rom gm@||@com
Thu Jan 17 19:28:46 CET 2019

  Jeff is correct, but maybe doesn't provide much context; the
documentation that he points doesn't contain details of the fitting
algorithm used in lme.  There is an entire book by Pinheiro and Bates
(Springer, 2000) that discusses the algorithms. The main thing to note,
though, is that lme uses a penalized least squares algorithm rather than
a moment-based/sum-of-squares decomposition, so technically there *are*
no "sums of squares" involved in the calculation.  For many problems
there are *equivalent* formulations in terms of sums of squares, and one
can often derive the SS involved, but lme doesn't actually use them.

On 2019-01-14 4:59 p.m., RICARDO ALVARADO BARRANTES wrote:
> I have a question related to the funcion LME in the library NLME.  I
> would like to understand how the F values are calculated, since the
> output only shows the degrees of freedom but doen't show the sums of
> squares involved in those calculations. 
> Thanks for your attention. 
> Ricardo
> 	[[alternative HTML version deleted]]
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

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