Levels of Measurement and Psychological Theory

Abstract

PSYCHOLOGISTS have recently displayed in creasing interest in fundamental problems of meas urement theory. In particular, they have shown con cern with the relevance of their measurement oper ations to the legitimacy of various statistical opera tions. Stevens' (10, 11) classification of measure ment levels into nominal, ordinal, interval and ra tio has served as a framework for most of the dis cussion. Unfortunately, certain assumptions under lying this classification scheme have been generally accepted without question. In Stevens' view, most of the devices that pass for measures in psychology should be regarded as indicants bearing an unknown, though probably monotonie, relation to the underly ing phenomena that we actually wish to measure. The indicant may be regarded as a measure when it is so calibrated that it yields scale values strictly isomorphic with quantities in the underlying phe nomenon.