Key points are not available for this paper at this time.
Thii paper presents an approach to the optimum design of a primary screening test for drugs. This approach takes into account the quantities which are at stake, i.e. the value of a successful outcome, the probability of a successful outcome and the effect on this of competition in research, the resources available, and the costs of the compounds and tests. By means of a numerical illustration it is shown that the optimum procedure is very robust to large errors in the estimation of all the input values; and that the optimum is very flat so that a wide range of procedures give near optimum results. This confirms and extends conclusions given in previous publications. The design of screening tests is related to the broader problems of project selection and the balance of the various activities in a research department.
O. L. Davies (1963) studied this question.