On Sat, 14 Dec 2002, Ludwik G�rski wrote, inter alia:
> ... if anyone knows anything about the Rank, please help me!
> Ludwik
Perhaps an example will help, with a few comments about the terms used.
Given a set of numbers, one can place them in order from smallest to
largest (they are then said to be in "ascending order"), or from largest
to smallest (this is "descending order"). Suppose there are N such
numbers in the set. One can then assign the numbers from 1 to N (as a
new variable) to the original values; these numbers (from 1 to N) are
called the "ranks" of the original values.
Thus, for the six values 17, 25, 7, 84, 46, 13
we can order them thus: 7, 13, 17, 25, 46, 84 (ascending order),
or thus: 84, 46, 25, 17, 13, 7 (descending order).
Their ranks would be 1, 2, 3, 4, 5, 6
and rank 1 could refer either to the largest or the smallest of the
original values, depending on how one chose to rank them, descending or
ascending respectively.
(I don't know Polish, but I think the German verb "sich reihen" means
"to arrange in order", or "to rank", if that's any help.)
If you have access to any of the statistical packages (SPSS, Minitab,
SAS, etc.), it will have a subroutine that produces ranks on request
(as Arthur Kendall's reply to you indicated, for SPSS).
If you don't, your computer surely has an ordering algorithm that will
put all the values in ascending or descending order, and you can then
assign rank numbers accordingly.
The main difficulty is that there may be ties (that is, several cases
that have the same value) in the original data, and you then have to
decide (1) how to notice them, to begin with, and (2) how to treat them
when you've found them. As Professor Kendall remarked, there are
several ways of doing this; for the application you mentioned in your
earlier post of December 13 ("problem with an algorithm"), I suspect
that taking the mean rank for tied values would work well. (This is
also, I believe, the default used by SPSS and Minitab.)
HTH. If any of this is not clear, please ask me to clarify. -- DFB.
-----------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
[was: 184 Nashua Road, Bedford, NH 03110 (603) 471-7128]
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
