Developing Scales to Measure Consumer Dispositions: A New Approach Based on Predictive Validity

Abstract

In the decades since publication of Churchill (1979), scale development in marketing has migrated in the direction of shorter scales and much smaller item pools. However, neither a justification for this trend, nor its implications for scale development, has been laid out. This paper focuses on the implications for measurement of consumer dispositions, with skepticism toward advertising serving as the running example. It identifies problematic aspects of the now obsolescent call for large item pools of 100, and clarifies the potential negative effects of pursuing internal consistency or clean factor structure in isolation from considerations of predictive validity. To address these concerns, an alternative procedure for generating and pruning item pools is offered. The core of the new approach is to apply well-accepted tools for scale validation at the level of individual items, thus moving the consideration of predictive validity forward to the item pruning stage.

Keywords: scale development, scale reduction, dispositions, predictive validity, single-item vs. multi-item measures

JEL Classification: M31