We compare the variances of three estimators (for lognormal distribution)
X SRS , X RSS and
X for different sample size n = 2, 3, …,10. X RSS is more
efficient than X SRS , the ranked set BLUE estimator
X performs best among these
three estimators considered by us. It is evident from the last two columns of Table 2 that
the relative efficiency depends on sample size and increases with the increase of sample
size. Thus we conclude, for lognormal distribution, that the obvious design rule to follow
is to choose the largest practicable ranked set samples within the overall sampling effort
and use the ranked set BLUE.
Log X