Buying Software with Exact Confidence.

We derive several new results which may be helpful to the buyers of software, testing for faults, or to the buyers of large lots, screening for defectives. Specifically, suppose that a fixed but unknown number n of faults or defectives remain before testing. In the testing phase they are observed at random times, T sub 1, T sub 2, ..., which are independent and identically distributed. Since testing is usually an on-going activity, this distribution is typically known. Under this assumption we derive a stopping criterion that guarantees that the tested software has no more than m defectives remaining with an exact probability 1 - alpha for any prespecified level alpha and integer m =n. We investigate its various properties, finite and asymptotic, and show that the rule is optimal in a broad sense. We also modify another conservative stopping rule investigated by Marcus and Blumenthal (1974) to make it exact and give numerical comparisons among the rules.