# Test if two values of Kendall's Tau are significantly different?

I want to use Kendall's Tau to check rank correlation in some data. I have two sets of data, and hence compute two values, $$\tau_a$$ and $$\tau_b$$. How do I test if $$\tau_a$$ is significantly different from $$\tau_b$$?

For perspective, If I were using Pearson/Spearman correlations, I would do a Fisher transformation followed by a Z-test to obtain a p-value. I something similar possible for Kendall's tau?

• – Glen_b Aug 18 at 11:56