A Bayesian Approach to Approximate Conditional Inference
本文扩展了Sweeting(1995)的统一框架，将贝叶斯方法用于推导近似条件推断公式，展示了如何利用贝叶斯论证得到Barndorff-Nielsen的公式，对统计推断研究者有参考价值。
Sweeting (1995) studies regular Bayesian and frequentist approximations within a unified framework in the case of a single parameter, and shows that higher-order approximations to sampling distributions arise from their Bayesian counterparts via an unsmoothing argument. In the present paper we extend this programme to include formulae in approximate conditional inference. In particular it is shown how Bayesian arguments may be used to derive some formulae developed by Barndorff-Nielsen (1980, 1983, 1986). The development proceeds in terms of likelihood roots.