> I think you should be using the Fisher transform
Thanks. I'll do this.
> though you should also be correcting for temporal autocorrelation
> when calculating the true temporal degrees of freedom.
The Fisher Transform is:
z = 0.5 * log[ (1+r)/(1-r) ]
where log = natural log.
There are no degrees of freedom in the Fisher transform?
If I am missing something here, then how would I correct for temporal autocorrelation?
Thanks,
- BettyAnn
|