Search

In this paper the authors propose some nonparametric selection procedures based on one-sample Hodges-Lehmann type estimators of the parameters for the symmetric location parameter populations. Some investigations of the efficiency are also discussed. (Author)

In sequential selection and ranking procedures, contributions have been made to select the 'best' population by using the indifference zone approach. But, very little work has been done for the subset selection problem. In this paper, the authors are interested in deriving some methods to construct nonelimination and elimination type sequential procedures to select a subset containing all 'mildly t best' populations, t fixed, 1 = or < t < k. A Monte Carlo study is carried out to investigate the efficiency of the elimination type procedures and tables are given at the end of the paper. (Author)

Let (Pi sub 1),...,(Pi sub k) be k independent populations with absolutely continuous distribution functions F sub (theta sub i) (x), i=1,...,k, where (theta sub i) are unknown real-valued parameters. The populations are ranked according to the values of the (theta sub i). The problem is to design a procedure to select a nonempty subset of all k fectorial rankings so as to include the correct ranking with a minimum probability P*. Section 2 considers the cases where the distribution of a statistic (T sub i) = T(X sub i1),...,(X sub in) from (Pi sub i) involves (theta sub i) as a location or scale parameter. Applications to the ranking of normal means (known common variance) and normal variances are discussed. (Modified author abstract)