CMStatistics 2016: Start Registration
View Submission - CFE
Title: Measurement error and rank correlation Authors:  Toru Kitagawa - University College London (United Kingdom) [presenting]
Jan Stuhler - Universidad de Madrid Carlos III (Spain)
Martin Nybom - SOFI - University of Stockholm (Sweden)
Abstract: A method is proposed to correct measurement error biases in the estimation of rank correlation coefficients (Spearman's rho and Kendall's tau). We approximate the measurement error biases using the technique of small error variance approximation developed and construct a bias-corrected estimator by subtracting the bias estimate from the sample rank correlations. Using the rich Swedish data in which the underlying precise measurements are available, we empirically assess approximation performances of our method in the two empirical applications; rank correlations of father and son's permanent income and father and son's cognitive abilities.