A Measure of Variability for the Customer Satisfaction Index Valentini P. DMQTE – Università G. D’annunzio Chieti-Pescara pvalent@unich.it Gattone S.A. DMQTE – Università G. D’annunzio Chieti-Pescara sgattone@yahoo.com Di Battista T. DMQTE – Università G. D’annunzio Chieti-Pescara dibattis@unich.it Abstract: In this paper we deal with the problem of identifying heterogeneity indices for the purpose of improving the analysis of customer satisfaction observing the phenomenon through a new perspective. Our work introduces some indices that may be used for measuring heterogeneity in Customer Satisfaction framework and an application on real data is illustrated. Keywords: heterogeneity, response pattern, intrinsic heterogeneity profile 1. Introduction Customer satisfaction (CS) is a central concept in marketing and is adopted as an important outcome measure of service quality by service industries. Most service companies have research programs designed to measure service quality and/or customer satisfaction. Such programs are designed to provide essential information to guide efforts to reduce variability in service quality. As highlighted by Giancristofaro et al. (2007), the concept of quality is strictly related to variability. This paper deals with the problem of identifying heterogeneity indices for the purpose of improving the analysis of customer satisfaction observing the phenomenon through a new perspective. In particular, in Section 2, we introduce the basic ideas, several indices and methods that may be used for measuring heterogeneity in Customer Satisfaction framework, while Section 3 deals with a case study in which the diversity profiles are used to mark differences among two subpopulations. 2. Defining and measuring customer satisfaction heterogeneity The semantic of a few terms used throughout this paper are addressed. The term response patterns is a convenient label for a set of distinct responses of a subject to a set of items (e.g. 5124 is a response pattern for four items coded from 1 to 5). Suppose we have an n by p data matrix of values of p ordered categorical variables, x 1 , x 2 ,…, x p with m i categories (i=1, 2,…, p) for n individuals. Any row of the data matrix is referred to as a response pattern and in general there are m 1 ×m 2 ×…×m p possible response patterns. If the sample size is much larger than m 1 ×m 2 ×…×m p , many of the response patterns will be repeated and they may be summarized by a matrix as a list of the observed response patterns together with their associated frequencies. Furthermore, the terms of heterogeneity and homogeneity need to be introduced even if they are concepts widely used in the field of statistics. The concept of homogeneity addresses the case in which every unit belonging to a population manifests the same category with respect to a statistical variable X. If this does not occur then heterogeneity is indicated by absence of homogeneity. Therefore the degree of heterogeneity obviously depends on the number of categories observed as well as on their associated frequencies. In particular the heterogeneity is at a minimum if the MTISD 2008 – Methods, Models and Information Technologies for Decision Support Systems Università del Salento, Lecce, 1820 September 2008 ___________________________________________________________________________________________________________ © 2008 University of Salento - SIBA http://siba2.unile.it/ese 115