In article <9gmcaa$75i$[EMAIL PROTECTED]>, "Glen Barnett"
<[EMAIL PROTECTED]> wrote:
> Monica De Stefani <[EMAIL PROTECTED]> wrote in message
> [EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > 2) Can Kendall discover nonlinear dependence?
>
> He used to be able to, but he died.
>
> (Look at how Kendall's tau is calculated. Notice that it is
> not affected by any monotonic increasing transformation. So
> Kendall's tau measures monotonic association - the tendency
> of two variables to be in the same order.)
>
> Glen
I do not understand why Kendall's Tau is being used instead of the
ordinary correlation coefficient with its partials and semi-partials.
For example, say you're reporting the correlation of X and Y (ice cream
consumed and water consumed) but Z (air temperature) might be secretly
responsible. So you correlate X and Z to find their linear dependence and
stash the residuals temporarily. Then you correlate Y and Z to find their
linear dependence--and stash the residuals again. Now you can revisit the
dependence of X and Y by correlating the residuals of X and Z versus the
residuals of Y and Z. The effect of Z has been partialed out.
You could try this using Kendall's Tau verus the ordinary correlation
coefficient to see if there is a difference. I personally have not run
into a data set where was a difference. BTW, significance tests are not
involved.
Doc
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
