The Continuous Multivariate Location-Scale Model Revisited: A Tale of Robustness
研究了所有多元连续抽样分布的位置-尺度族，发现固定单位球上特定分布并采用尺度参数的无信息先验时，位置和可观测量的推断相同，这解释了多种模型中的稳健性结果。
SUMMARY We consider the location-scale family of all multivariate continuous sampling distributions. Any model from the subclass induced by fixing a particular distribution on the unit sphere, combined with a standard improper prior on the scale parameter, will lead to the same inference on location and observables. This finding sheds light on many robustness results in less general models and its implications are illustrated in a number of examples.