Objective Bayesian testing for the linear combinations of normal means (Q2633420)

Assume that there are \(K\) independent and random samples from normal distributions with different means but the same variance. This paper is concerned with a problem to test whether a linear combination of the means is equal to a specific but known value. An objective hypothesis testing approach based on Bayes factors has been considered for this problem. This approach requires selection of appropriate priors. Two different priors, divergence-based and intrinsic, are considered in this paper. Formulas for the Bayes factors based on these priors are derived in the paper and algorithms based on respectively Markov chain Monte Carlo and one-dimensional numerical integrations are presented which can be used to calculate them for practical problems. Comparisons of the approaches based on these two priors are made through simulations and practical examples.