INCOMPLETE SAMPLING FOR SELECTION OF A FIXED PROPORTION

Testing only a sample of a population from which a fixed proportion are to be chosen for further use is considered in relation to the balance between testing costs and extra value achieved by culling the poorest members of the population. It is shown that in some circumstances the greatest return is obtained by testing some, but not all, members of the population. Optimum sampling fractions are deducted for selection on a quantitative chaaracter having a normal distribution, and approximate results are given for selection on a qualitative character.