The Structure of Customer Satisfaction: Effects of Survey Measurement

Abstract

This paper uses ideas from three different research streams to develop a model for customer satisfaction which explicitly allows for the context in which satisfaction is assessed. First, latent class/measurement error models are use to separate a latent variable of satisfaction from its indicators. Then, we borrow the survey researcher's notions of questionnaire response effects to allow direct interrelations among survey responses. Finally, a classic model of customer satisfaction, as adapted by Bolton and Drew (1991) is used to describe the form of the relationship among indicators, particularly those proximate items in a questionnaire.