On the Uniform Approximation of Laplace's Prior by t-Priors in Location Problems

Authors: 
Saurabh Mukhopadhyay, Malay Ghosh
Duke University, University of Florida

Nov 30 1992

The problem of deriving robust and classically acceptable Bayesian inference on location parameters is considered in this paper. The main result of the paper allows one to obtain uniform posterior approximation starting with a likelihood functions with heavy tails, e.g., t-distributions with unknown locations. The approximations of the relevant Bayes quantities are obtained as well.

Manuscript: 

PDF icon 1993-21.pdf