Selection Procedures for Order Statistics in Empirical Economic Studies

In a presentation to the American Economics Association, McCloskey (1998) argued that "statistical significance is bankrupt" and that economists' time would be "better spent on finding out How Big Is Big". This brief survey is devoted to methods of determining "How Big Is Big". It is concerned with a rich body of literature called selection procedures, which are statistical methods that allow inference on order statistics and which enable empiricists to attach confidence levels to statements about the relative magnitudes of population parameters (i.e. How Big Is Big). Despite their prolonged existence and common use in other fields, selection procedures have gone relatively unnoticed in the field of economics, and, perhaps, their use is long overdue. The purpose of this paper is to provide a brief survey of selection procedures as an introduction to economists and econometricians and to illustrate their use in economics by discussing a few potential applications. Both simulated and empirical examples are provided.