Document Server@UHasselt >
Education >
Archive >
PhD theses >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1942/8885

Title: Customer Satisfaction and Loyalty Research: a Bayesian Network Approach
Authors: Jaronski, Waldemar
Advisors: Vanhoof, Koen
Issue Date: 2004
Publisher: UHasselt Diepenbeek
Abstract: ... The unique contribution of this work comes mainly from the intersection of the Bayesian network Literature and the marketing modelling literature. In spite of their apparently attractive features for solving various marketing problems, Bayesian networks are, to the best of our knowledge, still not a well-recognized technique within the marketing community [Lilien and Rangaswamy, 2000, p.232]. This lack of recognition can be attributed to the following general reasons. Most importantly, we acknowledge that the methodology is still in the early stages of its maturity with respect to specific requirements of marketing research. In the context of causal modelling, some authors make even a parallel between the current stage of development of Bayesian networks with the stage of structural equation models in the 1970s [Anderson and Lenz, 2001]. Secondly, even taking into account its relative immaturity their use is appealing; however, there is a lack of a thorough discussion of basic features and potential added value of the Bayesian network technology as a tool in the arsenal a marketing researcher. This thesis is also motivated with the observation that little attention has been paid to date on adapting or evaluating Bayesian networks as a potential technique for conducting research, let alone marketing research. Instead, since its bloom in the 1990's, the vast majority of research on Bayesian networks has been focused rather on developing algorithms and fostering technical innovations for the purpose of expert systems. As such, this previous work has been limited to problems existing in artificial intelligence and data mining. Since we find it very important to bring the Bayesian network approach closer to marketing, as the overall goal of this thesis we aim to provide a critical evaluation of the application of Bayesian networks in theoretical and practical marketing research, and propose new methods and developments within the Bayesian network modelling to improve its current abilities with respect to specific requirements existing in the marketing research. However, this formulation of the overall objective would require an immense, if not unfeasible, task due to plethora of avenues in marketing research; therefore, we constrain ourselves to only one particular area in marketing science: the Customer Satisfaction and Loyalty (CS&L) research. Due to the growing importance of e-commerce and Internet in marketing science [e.g., Mahajan and Venkatesh, 2000; O'Connor and Galvin, 2001], we will consider the CS&L phenomenon both in the traditional, "mortar-and-brick" context as well as in the on-line one. Furthermore, the critical evaluation that we undertake in this thesis should be regarded as internal validation rather than external one. In other words, it is our aim to examine Bayesian networks individually rather than to compare this methodology in a competitive setting with other techniques applied today in CS&L research in terms of their respective outcomes and findings and to establish which techniques are superior and which perform worse. Consequently, we take the position by which the Bayesian network approach is considered in this thesis merely as another approach that can help understand and research the CS&L phenomenon. In order to achieve the aforementioned overall goal of the dissertation, it is also essential that the perspective we take here be from the position of a CS&L scientist rather than of a Bayesian network expert. In other words, we will take the needs and objectives in CS&L research as the starting point for this discussion. Consequently, let us now present our diagnosis of the requirements existing today in CS&L research and mark in more detail the areas in which the Bayesian network literature is still missing. ...
URI: http://hdl.handle.net/1942/8885
Category: T1
Type: Theses and Dissertations
Appears in Collections: PhD theses
Research publications

Files in This Item:

Description SizeFormat
Part 112.25 MBAdobe PDF
Part 213.11 MBAdobe PDF
Part 324.15 MBAdobe PDF

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.