Efficiency-Robust Systematic Linear Estimators of Location

Abstract

Several classes of efficiency-robust linear unbiased estimators of location based on k sample quantiles are defined. The efficiencies of these estimators under the normal, logistic, and double exponential distributions are computed for sample sizes n = 20, 30, 40, 50, 100, ∞, and subsamples of size k = 2, 6, 10, 20. These results are then discussed in relation to other research in efficiency-robust estimation.