Abstract
The purpose of this paper is to demonstrate the use of the super-population approach to the measurement and analysis of customer satisfaction on a national level. Since the estimation is based on a probability sample of customers and since the measurement model includes random errors, both sampling and measurement errors must be taken into account and calculated. The superpopulation approach for estimating the average customer satisfaction of a nation from a probability sample is described. The Horvitz-Thompson estimator is used for illustration. The total variance of the estimator is shown to consist of two components, one due mainly to the measurement process and the other due mainly to the sampling process. A numerical example from Statistics Sweden is described.