On selecting the best contender

摘要

When selecting the best contender based on relative rankings of k contenders, the definition of the best contender is not necessarily straightforward. In particular, if different summary statistics lead to selecting different best contenders, which summary statistic is to be relied upon is not usually clear. In this paper we define the best contender, discuss procedures compatible with the definitions, and compare the procedures via relative efficiency. An example data set is presented and discussed.