ABSTRACT
C. Gutenbrunner and J. Jureckova provided a crucial link between quantile regression ideas and rank tests as exposited by Hajek and Sidak through the formal duality of the linear programming approach. The new manuscript contained an extended introduction in which the reader tried to motivate the idea of L-statistics for regression along the lines of the work in the late 1940s and early 1950s by Mosteller and others. Gib was working with Lester Taylor on theoretical aspects of l1 $ l_1 $ regression, and Roger on hierarchical models for longitudinal data. Gutenbrunner and Jureckova provided a crucial link between quantile regression ideas and rank tests as exposited by Hajek and Sidak through the formal duality of the linear programming approach. Ruppert and Carroll, using earlier work by Jana Jureckova, also provided a much more straightforward proof of the asymptotic normality of the regression quantiles than our density-based approach.
