Course contentsShow
← Data Science
Lesson 489 of 2,14513. Correlation AnalysisFree lesson

Kendall's Tau Correlation Coefficient

Measuring concordance between paired observations using concordant and discordant pairs.

Kendall's Tau Correlation Coefficient

What you'll learn: How to measure the strength of association between two variables by counting concordant and discordant pairs.

What Makes Kendall's Tau Different

After learning about Spearman's rank correlation, you now have another rank-based tool: Kendall's Tau (τ). While Spearman measures monotonic relationships through ranks, Kendall's Tau takes a different approach—it measures concordance by examining every possible pair of observations.

Understanding Concordance

Imagine you have paired observations: heights and weights of five people. Pick any two people from your dataset. If person A is taller than person B and person A also weighs more than person B, that's a concordant pair—both variables agree on the ordering.

If person A is taller but weighs less, that's a discordant pair—the variables disagree.

The Logic Behind Tau

Kendall's Tau works by:

  1. Comparing every possible pair of observations
  2. Counting how many are concordant (C)
  3. Counting how many are discordant (D)
  4. Computing: τ = (C - D) / (total pairs)

If most pairs are concordant, τ approaches +1 (strong positive association). If most are discordant, τ approaches -1 (strong negative association). If concordant and discordant pairs roughly balance out, τ is near 0.

Why Use Kendall's Tau?

Kendall's Tau is especially useful when you have small sample sizes or many tied ranks. It has a more direct probabilistic interpretation than Spearman: τ represents the difference between the probability of concordance and discordance.

Key Takeaway: Kendall's Tau measures correlation by counting concordant and discordant pairs—it's a robust, intuitive alternative to Spearman that works well with small samples and ties.