🚀 go-pugleaf

In article <19s96oINNdd9@agate.berkeley.edu>, roberts@garnet.berkeley.edu (Seth Roberts) writes:
> I'm curious if anyone out there has made much use of robust
> location estimators, such as the trimmed mean and the biweight.
> And, if so, what happened?
>  
[deletion]
>  
> I'm curious why robust location estimators haven't become more popular.
[description of animal research deleted]
>  
> Seth Roberts
> Department of Psychology
> University of California at Berkeley

I believe that the reason robust location estimators
havn't taken off is that applied statisticians aren't
all that interested in location estimators per se - we
want to build models. The thrust of developments is
towards generalizing the linear model to non-linear,
non-normal, and non-iid situations. We'll do it
robustly if we know how, but if we don't we'll do it
anyway and patch things up with diagnostic statistics.
-- 
Murray A. Jorgensen [ maj@waikato.ac.nz ]    University of Waikato
Department of Mathematics and Statistics     Hamilton, New Zealand      
__________________________________________________________________
'Tis the song of the Jubjub! the proof is complete,
if only I've stated it thrice.'