A Methodology for Identifying and Testing Response Functions in Marketing with an Application to Customer Satisfaction

Marketing researchers are frequently confronted with a variety of competing models of the relationship between two variables, with each supported by its own theory and empirical findings. For example, in the area of customer satisfaction, a rich literature with strong behavioral underpinnings supported by extensive empirical studies has resulted in a series of models for the relationship between disconfirmation of expectations and customer satisfaction. Each model has different research and managerial implications, suggesting the need for a technique which can be used to explore the relationship and to compare the fit and predictive ability of competing models in a given situation. In the absence of contextual variables to explain when each of the different models should be used, we argue that an exploratory data analysis stage is necessary to guarantee an appropriate model for a particular application. We show how regression splines can be used to develop a flexible model for this purpose. Advantages of regression splines are that they ensure that the resulting model is consistent with the data, they provide a benchmark or null model against which specific hypotheses of theory-driven models can be tested, and they form a platform from which to learn more about the nature of the relationship being studied, including how it varies by market segment. After outlining the theory of linear splines and the methodology associated with their application, we demonstrate their flexibility to parsimoniously represent different functional forms using simulation data. We then apply the technique in the field of customer satisfaction to study the relationship between satisfaction and disconfirmation of expectations.