Munich Personal RePEc Archive

Robustness of Bayesian results for Inverse Gaussian distribution under ML-II epsilon-contaminated and Edgeworth Series class of prior distributions

Sinha, Pankaj and Jayaraman, Prabha (2009): Robustness of Bayesian results for Inverse Gaussian distribution under ML-II epsilon-contaminated and Edgeworth Series class of prior distributions.

[img]
Preview
PDF
MPRA_paper_15396.pdf

Download (342kB) | Preview

Abstract

This paper aims to study the sensitivity of Bayes estimate of location parameter of an Inverse Gaussian (IG) distribution to misspecification in the prior distribution. It also studies the effect of misspecification of the prior distribution on two-sided predictive limits for a future observation from IG population. Two prior distributions, a class ML-II ε-contaminated and Edgeworth Series (ESD), are employed for the location parameter of an IG distribution, to investigate the effect of misspecification in the priors. The numerical illustrations suggest that moderate amount of misspecification in prior distributions belonging to the class of ML-II ε-contaminated and ESD does not affect the Bayesian results.

UB_LMU-Logo
MPRA is a RePEc service hosted by
the Munich University Library in Germany.