Adaptive Robust Procedures for the One-Sample Location Problem

A family of rank statistics based on a set of simple scores functions is introduced. The distribution theory is briefly discussed, and both point and interval estimators for the center of an absolutely continuous symmetric distribution are presented. Given a random sample from a symmetric distribution, all subsequent inference is based on the member of the family of rank statistics having the smallest estimated asymptotic variance. The robustness properties are examined using the results of a simulation study.