Home › Stats › Hypothesis tests

KendallTau

Kendall's tau: do two rankings agree, pair by pair?

Where Spearman correlates the ranks themselves, this counts the pairs the two samples order the same way against the pairs they order oppositely, so it reads as a probability rather than as a correlation and is the steadier of the two on a short sample. Both normalisations scipy publishes are offered — see KendallVariant — and they share a p-value, because the count they normalise is the same count.

Members

Member	What it does
KendallTau.Test	Correlates two paired samples by counting concordant and discordant pairs.