Mixed group validation: A method for determining the validity of diagnostic signs without using criterion groups.

To determine whether a sign x distinguishes between X-type individuals and Y-type individuals (Y = not-X), it is necessary to obtain estimates of p(x/X) (the probability an X-type shows sign x) and p(x/Y). The traditional method for determining these probabilities, criterion-group validation, involves observing the incidence of x in one group consisting entirely of X-types and in a second group consisting entirely of Y-types. Here, equations are developed for obtaining these probabilities by observing the incidence of x in any two groups having different, known base rates of X-types. The validation procedure making use of these equations is termed mixed group validation, and it is pointed out that criterion-group validation is a special (limiting) case of mixed group validation. Advantages of mixed group validation are discussed.