A Study on Nonparametric Selection Procedures for Scale Parameters

  • Song, Moon-Sup (Department of Computer Science and Statistics, Seoul National University, Seoul 151) ;
  • Chung, Han-Young (Department of Computer Science and Statistics, Seoul National University, Seoul 151) ;
  • Kim, Dong-Jae (Department of Computer Science and Statistics, Seoul National University, Seoul 151)
  • Published : 1985.06.01

Abstract

In this paper, we propose some nonparametric subset selection procedures for scale parameters based on rank-likes. The proposed procedures are compared to the Gupta-Sobel's parametric prcedure through a small-sample Monte Carlo study. The results show that the nonparametric procedures are quite robust for heavy-tailed distributions, but they have somewhat low efficiencies.

Keywords