Research in focus: Measurement in marketing: current scenario, recommendations and challenges.
An assessment of measurement best practices in marketing, an area in which the great challenge is to attribute values to concepts that are not directly observable (such as satisfaction, loyalty, happiness and attitude toward a brand).
Objective: To provide a comprehensive summary of the measurement process in marketing by presenting classic models and their variations as well as alternatives and trends.
• Presentation of the classic models for measuring underlying constructs as well as the limitations and critiques of these models
• Reflection on the measurement process in cross-cultural studies
• Discussion of the trends in the measurement process in marketing
• Classic measurement models in marketing have unquestionable merit because they are the most appropriate for constructs that have multiple and reflective indicators. They also emphasize predominantly quantitative aspects in the validation process.
• There are alternatives and complements to the classic measurement model; these are applied above all to single and formative indicators. (In these indicators, an underlying factor explaining the variation in a set of indicators is not assumed but is thought of inversely. In other words, the variation in items implies variation in the construct that is formed.) Further, these alternative models are supported by qualitative validation procedures, which are commonly ignored in research in the area.
• Of all the trends in quantitative measurement, the most prominently used techniques are Bayesian estimation, item response theory (IRT) and partial least squares (PLS) models.
• The study suggests alternatives to the classic model for those types of measurement for which it is less appropriate, in particular, known group validity procedures, numbers of points on scales and aggregation strategies.
• Qualitative validation procedures, such as content and face validity, deserve greater attention and should be prominently used in marketing measurement.
• Because of its properties, item response theory (IRT) tends to produce more stable results in the validation of measures developed in cross-cultural contexts than classic procedures, such as factor analysis and structural equation modeling.
• It is recommended that marketing researchers dedicate time to ensure that their measurement models are perceived as strong. Only subsequently does it make sense to develop advanced models for testing hypotheses between constructs.