Re: st: RE: Signtest (statistics question), thanks and a further question

[Roger Newson <roger.newson@kcl.ac.uk>]

At 16:04 12/05/03 +0000, Guillaume Frechett wrote:
> Thanks to the three guys that responded. Nick Cox wrote:
>
> > -ranksum- works quite well in this situation.
> My understanding (which I think is confirmed by Roger Newson's answer -- 
> see below) was that since I have paired observations Ranksum would be 
> inapropriate.
>
> Roger, should I understand from your answer that there are tests taking 
> into consideration what I am talking about but that 1) I have to read your 
> thesis to learn about it, and 2) that it wouldn't be trivial to code? We 
> can move this discussion to direct email if that is more appropriate.

In the paired case, you do not have to read my thesis to learn about the 
test to use. You simply type

gene sgndif=sign(x1-x2)
ci sgndif

and you get a CI for the unconditional mean signed difference 
(corresponding to Kendall's tau-a). If you then type

ci sgndif if sgndif!=0

then you get a CI for the conditional mean signed difference (corresponding 
to Somers' D). My thesis extended the principle to groups larger than 
pairs, eg extended families of yeast cells or mouse embryo cells.

I can send Guillaume a pre-publication draft of my Stata Journal insert if 
he wants. It includes an example of using -ranksum- and -somersd- to 
compare 2 independent groups, ie American cars and non-American cars. As 
Guillaume says, -ranksum- is not appropriate in the paired case.

Best wishes

Roger


--
Roger Newson
Lecturer in Medical Statistics
Department of Public Health Sciences
King's College London
5th Floor, Capital House
42 Weston Street
London SE1 3QD
United Kingdom

Tel: 020 7848 6648 International +44 20 7848 6648
Fax: 020 7848 6620 International +44 20 7848 6620
  or 020 7848 6605 International +44 20 7848 6605
Email: roger.newson@kcl.ac.uk

Opinions expressed are those of the author, not the institution.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/