Restrictions on statistics imposed by method of measurement: Some reality, much mythology

Abstract

Practically everyone who has been exposed to a course or a textbook covering research methods in the social sciences realizes that there are four scales of measurement widely referred to in the field: nominal, ordinal, interval, and ratio. As a result of that exposure, he or she may be concerned about using appropriate statistics when a given scale is at hand. Some anxiety, for example, may result from the mere thought of using a t-test with ordinal data. This article discusses the origins and philosophical bases of arguments regarding scale-statistics relationships; points out misconceptions and fallacies in certain of these arguments; and emphasizes that in the overwhelming majority of cases in the social sciences, most particularly in criminology and criminal justice, researchers may use statistics without concern or anxiety regarding scale properties. © 1984.