R
Posts:
31
Registered:
10/8/10


Test on Difference in Mean Correlations
Posted:
Sep 5, 2012 10:08 PM


Hello:
Is there a way to perform a formal statistical test on whether two mean correlations [calculated from data from the same sample] are significantly different from each other:
H0: mean[corr(1,2), (corr 3,4)] = mean[corr(5,6), (corr 7,8)]
I realize that since correlations are not additive, in order to calculate the mean correlations one would likely use the Fisher's z transformation or perhaps use squared correlations, but I'm stuck after this point.
I realize this is a peculiar questionjust curious if hypothetically this would be possible, and if so, how to do it.
Thanks!

