Generalizability modeling of the foundations of customer delight

Purpose – This research seeks to present a methodology for investigating the generalizability of a theory‐testing model. The methodology is used to examine the generalizability of a model of the antecedents and consequences of customer delight. Design/methodology/approach – Theory testing of models in the marketing often fails to define an intended universe of generalization. This paper shows how multivariate generalizability theory can be used to estimate construct covariance components for specific sources of variance. These components can then be used to assess the generalizability of a structural equation model of a marketing phenomenon. Findings – The parameters of a model of customer delight obtained from data that sample customers of a service or data that confound sources of variance do not generalize to data that capture variation across services or variation across raters. The relative impact of customer delight and satisfaction on behavioral intention varies with the source of variation being studied. Practical implications – Previous research suggests that after controlling for customer satisfaction, customer delight accounts for very little variation in behavioral intention. But, for the source of variation of most relevance to managers, namely web sites, it is customer delight, not customer satisfaction, that is strongly associated with behavioral intention. Originality/value – The methodology can be applied and can produce model parameters having substantially different managerial implications for the management of customer satisfaction and customer delight.