Kendall's Tau
You'll learn an alternative measure of rank correlation that counts how often pairs of observations agree in their ordering.
What Is Kendall's Tau?
You've already learned Spearman's rank correlation, which adapts Pearson's formula to ranks. Kendall's Tau takes a completely different approach: it examines every possible pair of observations and counts whether they're concordant (rankings agree) or discordant (rankings disagree).
The Core Logic
Imagine you rank two variables for several observations—say, students ranked by study hours and by test scores. Pick any two students. If the student who studied more also scored higher, that pair is concordant. If the one who studied more scored lower, that pair is discordant.
Kendall's Tau is essentially:
τ = (number of concordant pairs − number of discordant pairs) / (total pairs)
More precisely:
τ = (C − D) / (n(n−1)/2)
where:
- C = concordant pairs
- D = discordant pairs
- n = number of observations
- The denominator is the total number of possible pairs
Why Use Kendall's Tau?
- Robust interpretation: Tau directly measures the probability that two randomly chosen pairs are in the same order
- Better for small samples than Spearman in some cases
- Handles ties more gracefully with adjusted formulas (Tau-b, Tau-c)
- Values range from −1 (perfect disagreement) to +1 (perfect agreement)
Kendall's Tau typically yields smaller absolute values than Spearman's ρ for the same data, but both detect the same monotonic relationships.
Key Takeaway: Kendall's Tau measures rank correlation by counting concordant and discordant observation pairs, offering a probability-based interpretation of monotonic association.