You are not currently logged in.
Access your personal account or get JSTOR access through your library or other institution:
If You Use a Screen ReaderThis content is available through Read Online (Free) program, which relies on page scans. Since scans are not currently available to screen readers, please contact JSTOR User Support for access. We'll provide a PDF copy for your screen reader.
On Some Problems in Bayesian Model Choice in Hydrology
L. R. Pericchi and I. Rodriguez-Iturbe
Journal of the Royal Statistical Society. Series D (The Statistician)
Vol. 32, No. 1/2, Proceedings of the 1982 I.O.S. Annual Conference on Practical Bayesian Statistics (Mar. - Jun., 1983), pp. 273-278
Stable URL: http://www.jstor.org/stable/2987625
Page Count: 6
Since scans are not currently available to screen readers, please contact JSTOR User Support for access. We'll provide a PDF copy for your screen reader.
Preview not available
Described is the Standard Bayesian Procedure (SBP) for model choice between K competing models as used in standard practice. The first problem which is discussed is the tendency of the SBP to yield lower posterior model probabilities for some correct sampling models, for low to moderate sample sizes, even when both models have equal numbers of adjustable parameters. This problem is exhibited in a case of model selection for hydrological extremes between Normal and Lognormal models. It is claimed in Pericchi (1981b), that the reason behind the problem is that the SBP tends to inflate the posterior probability of the model that has the smallest expected increase in information about its parameters. He then proposed to modify the SBP using an information criterion, that we adopt here. Finally, the discrimination between realistic models for hydrological extremes as Log-Pearson III, Gumbel, etc., is discussed. Previous approaches to incorporate prior information are analysed, and a tentative alternative procedure is described.
Journal of the Royal Statistical Society. Series D (The Statistician) © 1983 Royal Statistical Society