A Density-Quantile Function Approach to Choosing Order Statistics for the Estimation of Location and Scale Parameters.

It has been shown that the estimation of location and scale parameters by linear systematic statistics may be formulated as a problem in regression analysis of a smoothed sample quantile process. In this dissertation, a general approach to optimal spacings selection is presented that utilizes design techniques for continuous parameter time series regression. This methodology is applied to several common distributions. The problems of optimal order statistic selection for estimation in censored samples, for quantile estimation and for the summarization of large data sets are also considered.