Item Context and the Stability of Entity-based and Attribute-based Multiattribute Scaling Methods

Abstract

Multientity scaling is a multiattribute-measurement procedure that obtains simultaneous ratings of two or more entities from a single respondent. The purpose of this research is to examine experimentally the issue of item-context effects on the stability of two types of multientity scaling procedures—attribute-based and entity-based multientity scaling. The findings indicate that both procedures may be subject to problems of instability when the comparative context of the ratings is changed, with attribute-based scaling being characterized by more instability than entity-based scaling. On the basis of the findings, potential methods of increasing multientity scaling stability are explored.