Measuring marketing constructs : a comparison of three measurement theories

Abstract

A large number of new constructs are introduced into marketing. These new constructs are important in the development of marketing theories. The validation of these constructs are primarily based on a factor analytical framework (e.g., Churchill’s paradigm), with a validation rationale found in Classical Test Theory. However as the limitations of Classical Test Theory are widely realized by many researchers, alternative measurement theories might provide better and more coherent guidelines for marketing researchers with regard to how to measure and validate marketing constructs. In this thesis I look at two alternative measurement theories, Item Response Theory and Generalizability Theory, in addition to Classical Test Theory. Both Item Response Theory and Generalizability Theory have recently become more important for marketing measurement. This thesis addresses how the constructs are measured based on these three theories and how these measurement theories differ from each other. After a detailed review of each theory, the theories are contrasted, especially in terms of construct validation. It is found that Classical Test Theory, Item Response Theory, and Generalizability Theory vary in how they address different measurement issues. They differ in terms of how constructs are defined, measured, and validated. However, the validation process employed for these three theories can only provide empirical evidence, or indications, for the construct validity but cannot provide evidence as to whether constructs exist or not. This will be a challenging research question for future research.