Besides saying that a relation is positive or negative is it possible to interpret the Kendall's Tau-B depending on its level?
E.g.: I have two significant values 0.2 and 0.6, would be the interpretation of these values be different and if so why?
Solved – How to interpret the level of Kendall’s Tau-b
interpretationkendall-tau
Best Answer
I am not sure of what you mean exactly and what are your expectations.
Basically, $0.2$ and $0.6$ indicate both positive dependence/concordance, but $0.6$ indicates a stronger concordance between your variables.
I think copulas (cf. this article for a technical presentation of the links between dependence/concordance and copulas) are one of the good ways to visualize and understand this coefficient (btw, the rank correlations (spearman, kendall, and so on) can be defined in terms of the copula of the variables).
For instance, look at this scatterplot (probability integral transform / empirical copula transform) of your two variables:
In this case, they are more or less kendall-correlated with a strength of $0.6$. If it were $0.2$ the points-cloud in the scatterplot would be more uniform, it if were $0.99$ the points-cloud would be near to the straight line diagonal of $[0,1]^2$.