Abstract
Coefficients for computing ordered linear unbiased minimum variance (OLUMV) estimators of the location and scale parameters have been obtained for selected members of a class of symmetric power distributions which includes the normal, double exponential, and as a limiting case, the rectangular. Two examples are given illustrating how these coefficients may be used to study the sensitivity of inferences concerning location parameters to departure from normality. The first example involves a single location parameter while the second extends the method to linear contrasts of several location parameters. Other uses of the coefficients are indicated, and a comparison of the efficiencies of the OLUMV estimators with several commonly used estimators of location is given.

This publication has 0 references indexed in Scilit: