Presentation of search results: the<i>CL_s</i>technique

Abstract

I describe a framework for the presentation of search results which is motivated by frequentist statistics. The most well-known use of this framework is for the combined search for the Higgs boson at LEP. A toy neutrino oscillations experiment is used to illustrate the rich information available in the framework for exclusion and discovery. I argue that the so-called CL/sub s/ technique for setting limits is appropriate for determining exclusion intervals while the determination of confidence intervals advocated by Feldman and Cousins' method is more appropriate for treating established signals, i.e. going beyond discovery to measurement. (17 refs).