Conjoint Measurement and the Analysis of Noisy Data: A Comment

Abstract

In a previous study reported here Moriarity and Barron [1976] used conjoint measurement to model the materiality judgment process of audit partners. They did so following the approach suggested by Krantz and Tversky [1971] in which axiomatic conjoint measurement (ACM) is used to identify the functional form of the decision maker's judgment model, and then numerical conjoint measurement (NCM) is applied to find the best-fitting scale values (part-worth functions).' Shortly thereafter, the American Accounting Association's Committee on Human Information Processing (AAA [1978]) suggested that conjoint measurement would be useful in the study of certain aspects of human information processing in accounting research, particularly for testing alternative composition rules which uses only ordinal properties of the data [1978, p. 32]. Composition rules refer to the functional forms (e.g., additive, multiplicative, etc.) of decision makers' judgment models.2 In a later paper, Moriarity and Barron [1979] examined the preaudit materiality judgments of audit partners, in which they assumed an additive model and used a holistic orthogonal parameter estimation procedure (Barron and Person, [1979]). Swieringa [1979] criticized this