Please use this identifier to cite or link to this item:
Title: Application of Bayesian modeling to management information systems: A latent scores approach
Authors: Gupta, S. 
Kim, H.-W.
Issue Date: 2007
Citation: Gupta, S., Kim, H.-W. (2007). Application of Bayesian modeling to management information systems: A latent scores approach. Bayesian Network Technologies: Applications and Graphical Models : 103-126. ScholarBank@NUS Repository.
Abstract: This chapter deals with the application of Bayesian modeling as a management decision support tool for management information systems (MIS) managers. MIS managers have to deal with problems which require prediction and diagnosis for decision making. Lacking a proper tool for making informed decisions, MIS managers feel hard-pressed for a scenario analysis which can take into account the proper causal relationships existing in the real world. Bayesian modeling could be an appropriate support tool for such decision making. However, its application to decision support in MIS is different from application to other fields, as the variables in field of MIS are hypothetical. This brings in a need for Bayesian modeling at a hypothetical variable level rather than at the observed variable level. In this chapter we will study how Bayesian modeling can be used as a tool for managerial decision support in MIS. The conclusions of this chapter can also be extended to other social science researches where the variables are hypothetical in nature. Structural equation modeling (SEM) is good for empirical validation but it is not suitable for prediction and diagnosis. Prediction and diagnosis are useful for managerial decision support and can be done using Bayesian networks. Bayesian networks, however, do not differentiate between causal and spurious relationships. The capability of SEM in empirical validation combined with the prediction and diagnosis capabilities of Bayesian modeling offers an excellent tool for managerial decision support. This study proposes the linkage of SEM to Bayesian testing, for prediction and diagnosis from an empirically validated model. We apply the proposed approach to management decision support for customer retention in a virtual community. This research helps SEM researchers in extending their models for managerial prediction and diagnosis. It benefits Bayesian researchers by providing for the application of modeling causal relationships at a latent variable level. Modeling at the latent variable level, before Bayesian testing, would help in simplifying and uncovering the situation under study, and facilitating the identification of causal relationships. © 2007, IGI Global.
Source Title: Bayesian Network Technologies: Applications and Graphical Models
ISBN: 9781599041414
DOI: 10.4018/978-1-59904-141-4.ch006
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.


checked on Sep 20, 2020

Page view(s)

checked on Sep 18, 2020

Google ScholarTM



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