Abstract:
Educational and vocational psychologists are familiar with the fact that, having established a criterion of proficiency in a given subject area or job specialty, and having developed a set of reliable predictors, a number of problems arise which relate to the mathematical methods to be used in deciding from the information available which individuals are most likely to be successful. This problem becomes more acute when each individual can be assigned to one of a number of different activities, within each of which he has a different probability of future success, particularly if it is more important from the institutional standpoint, to approximate the maximum level of effectiveness in certain activities than in others. R. Von Mises, "On the classification of data into Distinct Groups". Ann Math.Stat. 16: pp68-73 (1945).
Although rigorous mathematical methods have been developed to determine predictor-criterion relationships, and to predict a given criterion with maximum accuracy in a least-squares sense, the mathematical models for effecting differential assignment are frequently so complex that the task is more often accomplished by rule-of-thumb procedures, or in terms of professional judgment for which there is a minimum of systematic rational support.
