Procedures for Simultaneous Estimation & Selection of the Best Population with a Preliminary Test

SYNOPTIC ABSTRACT

In this paper we study the problem of selecting the best population with the option not to select if the populations are tested to be not significantly different, and simultaneously providing an interval estimate for the parameter of a hest population whenever a best population is selected. A general formulation of the selection problem with emphasis on location and scale parameters is proposed, and its application to the normal distribution is given.

Key Words and Phrases:

• significant test
• normal distribution
• t distribution
• largest and smallest means
• largest and smallest scale parameters
• interval estimation