On scaling performance operating characteristics - Caveat emptor

Abstract

Problems associated with scaling and normalizing empirical performance operating characteristics (POCs) are examined. Normalization methods proposed by Wickens (1980) and by Mountford and North (1980) are critically evaluated. Computer simulations are used to generate raw-score and normalized POCs. The interpretation of transformed empirical POCs (Wickens, Mountford, and Schreiner, 1981) is shown to contain inconsistencies. The normalization techniques reviewed fail to resolve POC scaling problems. Caution must be exercised when interpreting transformed POCs.