This paper characterizes and proposes a method to correct for errors-in-variables biases in the estimation of rank correlation coeffcients (Spearman's ρ and Kendall's τ). We first investigate a set of suffcient conditions under which measurement errors bias the sample rank correlations...

This paper characterizes and proposes a method to correct for errors-in-variables biases in the estimation of rank correlation coeffcients (Spearman's ρ and Kendall's τ). We first investigate a set of suffcient conditions under which measurement errors bias the sample rank correlations toward...