Bayesian Model Averaging

Remarks from Literature:

‘The basic idea of Bayesian Model Averaging (BMA) is to make inferences based on a weighted average over model space.’ (Hoeting*)

‘BMA accounts for uncertainty in (an) estimated logistic regression by taking a weighted average of the maximum likelihood estimates produced by our models…and may be superior to stepwise regression’ ( Goenner and Pauls, 2006)

‘The uniform prior ( which assumes that each of the K models is equally likely)… is typical in cases that lack strong prior beliefs.’ (Goenner and Pauls, 2006 and Raftery 1995).


References:

Goenner , Cullen F. and Kenton Pauls. ‘A Predictive Model of Inquiry to Enrollment.’ Research in Higher Education, Vol 47, No 8, December 2006.

*Jennifer A. Hoeting, Colorado State Univeristy. ‘Methodology for Bayesian Model Averaging: An Update’

Hoeting, Madigan, Reftery, and Volinksy. ‘ Bayesian Model Averaging: A Tutorial.’ Statistical Science 1999, Vol 14, No 4, 382-417.

Peter Kennedy. ‘A Guide to Econometrics.’ 5th Edition. 2003 MIT Press

Raftery, Madian and Hoeting. ‘Bayesian Model Averaging for Linear Regression Models.’ Journal of the American Statistical Association (1997) 92, 179-191

Raftery A.E. ‘Bayesian Model Selection in Social Research .’ In: Marsden, P.V. (ed): Sociological Methodology 1995, Blackwells Publishers, Cambridge, MA, pp. 111-163.

Studenmund, A.H. Using Econometrics A Practical Guide. 4th Ed. Addison Wesley. 2001

SAS/STAT 9.2 User’s Guide. Introduction to Bayesian Analysis Procedures.