為了研究在某些混亂的集合中,察視保有固有的、健全的特性,所以我們將在這些種類之中,使用在兩個事後分配之間的局部發散性測度。對於事前機率(Pri or)或概似函數(Likelihood)而言,所獲得的局部發散性極限值來判定,其值愈小意味著愈有健全性。本研究考慮兩種的混亂(事前機率與概似函數),分析局部性敏感度。此外,我們亦考慮概似函數來自加權分配(Weighted distribution)的情況。最後,我們提出一些量化的例子,並加以說明。 This paper considers the use of local φ-divergence measures between posterior distributions under classes of perturbations in order to investigate the inherent robustness of certain classes. The smaller value of the limiting local φ-divergence, implies more robustness fo r the prior or the likelihood. In this paper, two kinds of the perturbations (prior and likelihood) are considered for the local sensitivity analysis. In addition, we also consider the cases when the likelihood comes from the class of weighted distributions. Finally some numerical examples are considered which provides measures of robustness.