Sensitivity analysis of blue for the population mean based on a ranked set sample

The use of best linear unbiased estimator (BLUE) has received much attention in estimating the population mean, based on a ranked set sample. However, the underlying distribution should be known for such estimation. This paper attempts to investigate whether the BLUE is sensitive to the misspec-ification of the underlying distribution. Various distributions are considered for comparisons of its performance with the classical ranked set sample mean. It is found that in general sensitivity of the relative performance of the BLUE mainly depends on the kurtosis but not the skewness of the underlying distribution. Copyright © 1998 Taylor and Francis.