Hi
You can use Spearman's test for independence which states that if the
two rankings are independent, then sqrt(n-1).r_S is assymptotically
normal(0,1). However, if in your application it makes more sense to give
more importance to some rankings, for instance the first ones, then you
should use a weighted measure (see for instance the article by Pinto da
Costa & Soares (2005) in the Australian and New Zealand Journal of
Statistics)
Regards
Joaquim Costa
On Mon, 2008-02-18 at 08:45 +0100, Alphonse Monkamg wrote:
> Dear ALL,
> Does any one know a test statistics on how to comparing two ranking methods.
> For instance, suppose we are ranking 8 students based on their performance on grade on Mathematics for sequences A and B. A gives ranking (1, 2,3,4,5,6,7,8) and B gives(2,1,4,3,5,7,6,8).
>
> How to test that these two ranking scheme are equal or not.
>
>
> Thanks in advance
>
> Alphonse
>
>
>
> ---------------------------------
> Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail
--
Joaquim Fernado Pinto da Costa
Departamento de Matemática Aplicada
Faculdade de Ciências da Universidade do Porto
Rua do Campo Alegre, 687
4169 - 007 Porto
Portugal
Tel: (+351) 220 402 270/202
Fax: (+351) 220 402 209
|