On 27 Dec 1999 03:14:17 -0800, [EMAIL PROTECTED] (Gerhard
Rumpold) wrote:
>
> Does anyone know about a method to compare two cohen-kappa values, if they
> are significant different from each other.
>
> I have a Kappa(t1) of r=.64
> and Kappa(t2) of r=.72 .
>
> How can I compute that at the time t2 the Kappa is a real improvment
> compared to t1.
to be a little more specific,
- so what is supposed to be greater than what? is it the correlation
of #1 or # 2 with #3 -- in that case, you want to list the individual
scores and see what you can say about who is closer to #3.
If you are comparing two different samples, then you need to keep in
mind that your statement is about the measures within those *samples*,
and those two numbers are not very different, speaking in general
about how much samples do differ.
Are the kappas 2x2? Kappa is not a very good statistic where the
marginals might differ.
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
