Title: Role of information in classical and Bayesian modelling

Authors: K. Surekha Rao, Omprakash K. Gupta

Addresses: School of Business and Economics, Indiana University Northwest, 3400 Broadway, Gary, IN 46408, USA. ' Department of Management and Marketing, P.O. Box 638, College of Business, Prairie View A&M University Prairie View, TX 77446-0638, USA

Abstract: Information plays a significant role in a decision-making process. Managerial decisions require applied statistical analysis. It is a common practice to use the method of classical least squares. It was observed that indirect least squares type estimation for structural models yields empirically unstable and highly variable results. In this paper we show that this instability and high variability of estimates stems from the lack of information in classical modelling. We suggest Bayesian modelling that allows decision makers to incorporate more information and yields stable and improved results. We provide empirical results in support of the role of information.

Keywords: least squares estimation; structural models; Fisher|s information; precision; Bayesian modelling; operational research.

DOI: 10.1504/IJOR.2007.014172

International Journal of Operational Research, 2007 Vol.2 No.4, pp.429 - 439

Published online: 24 Jun 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article