Approximate Inference in Location-Scale Regression Models
论文证明卡方分布和多元t分布能很好近似位置尺度回归模型的后验分布，对线性回归模型给出贝叶斯与非贝叶斯解释，并用威布尔回归和t误差回归验证。
Abstract It is shown that χ2 and multivariate t distributions provide good approximations to posterior distributions arising in location-scale regression models. In the case of a linear regression model, Bayesian inferences based on an improper prior have conditional frequency, as well as fiducial/structural interpretation; so the results are also applicable in a non-Bayesian context. Numerical support for the approximations is presented for the case of a Weibull regression model and a regression with t distributed errors.