Re: [R] outlier tests

From: Prof Brian Ripley <ripley_at_stats.ox.ac.uk>
Date: Thu 01 Jul 2004 - 01:12:59 EST


I thought outlier tests were mainly superseded two decades ago by the use of robust methods -- they certainly were in analytical chemistry, for example. All outlier tests are "bad" in the sense that outliers will damage the results long before they are detected. See e.g.

@Article{AMC.89a,

  author       = "{Analytical Methods Committee}",
  title        = "Robust statistics --- how not to reject outliers. {Part}
                  1. {Basic} concepts",
  journal      = "The Analyst",
  volume       = "114",
  pages        = "1693--1697",
  year         = "1989",

}

On Wed, 30 Jun 2004, Greg Tarpinian wrote:

> I have been learning about some outlier tests -- Dixon and Grubb,
> specifically -- for small data sets. When I try help.start() and search
> for outlier tests, the only response I manage to find is the Bonferroni
> test avaiable from the CAR package... are there any other packages the
> offer outlier tests?

That's not an outlier test in the sense used by Dixon and Grubb, but is an illustration of the point about robust methods being better, in this case protecting better against multiple outliers.

> Are the Dixon and Grubb tests "good" for small samples or are others
> more recommended?

-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

______________________________________________
R-help@stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Received on Thu Jul 01 01:17:47 2004

This archive was generated by hypermail 2.1.8 : Fri 18 Mar 2005 - 04:46:47 EST