Cues and Components in Multiattribute Evaluation

Abstract

Multiattribute additive evaluation models largely ignore the causal structure connecting attributes and overall evaluations. We propose here a distinction between two such causal structures. In "cue-type" tasks, attributes reflect or indicate the value of a distal variable measuring all or part of the entity′s worth. In "component-type" tasks, attributes are themselves the constitutive elements of worth. Attributes of the two kinds should be combined differently. Results are reported from two experiments in which the verbal context was manipulated so as to induce either cue-type or component-type interpretations of identical attribute sets. The data strongly suggest that subjects are sensitive to the distinction and respond in theoretically appropriate ways.