Hi Joe,
>> Also, since correlations have variable variance, you should use a Fisher's
>> Z transformation, as that will stabalize the variance.
>
> Hi Tom, what about covariances? Do covariances have variable variance? (I
> recently needed to do permutation tests on something more like covariances
> rather than correlations.)
This is an issue as well. Covariances have something like a scaled
chi-squared distribution, and so the variance of the covariance
estimate will increase the the true covariance.
Maybe you can instead separate out the inference to be first on
variance and then on correlation (upon which you can then use a
Fisher's transform).
-Tom
-- Thomas Nichols -------------------- Department of Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
[log in to unmask] 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029
|