Re: Summary of Robust Regression Algorithms

Eva Cantoni (
Tue, 06 Jan 1998 10:14:38 +0100

This is a multi-part message in MIME format.

Content-type: text/plain; charset=us-ascii
Content-transfer-encoding: QUOTED-PRINTABLE

David Ross wrote:

> (...) Packages (like S+), coupled with the ubiquity of remarkable c=
> power, sometimes make it a bit *too* tempting to drop data into rob=
> procedures. I'm guilty of this myself (especially since the robust=
> of lmsreg is backed up by a beautiful geometric characterization). =
> anticipation of Doug Martin's response, let me offer (to anyone rea=
> this thread) an example I sometimes give my students. Suppose your
> dataset has 2k+1 elements in it, k+1 of them take the value 0, and =
k of
> them take the value 100. The median is therefore 0. Now, change on=
e of
> the 0s into a 100. The median is now 100. This is robust?

Will the means at 100*k/(2k+1) and 100*(k+1)/(2k+1) best resume this =
data? I
don't think so.

This is not a problem of the median. There is no point looking for a =
mesure of a dichotomous variable !!!

Eva Cantoni

Content-type: text/x-vcard; charset=us-ascii; name=vcard.vcf
Content-description: Card for Eva Cantoni
Content-disposition: attachment; filename=vcard.vcf
Content-transfer-encoding: QUOTED-PRINTABLE

begin: vcard
fn: Eva Cantoni
n: Cantoni;Eva=20
org: D=E9part. d'Econom=E9trie, Universit=E9 de Gen=E8ve
adr;dom: 102, Bvd. Carl-Vogt;;;CH - 1211 Gen=E8ve 4 ;;;
tel;work: (+41) 22 705 8240
tel;fax: (+41) 22 705 8299=20
x-mozilla-cpt: ;0
x-mozilla-html: FALSE
version: 2.1
end: vcard