贝叶斯风险比的蒙特卡洛误差在对数二项回归模型中的估计：一种高效的MCMC方法
摘要：Binary outcomes in cohort studies are often analyzed using logistic regression. However, it is known that logistic regression is not suitable for estimating risk ratios when the outcome is common. In such cases, a log-binomial regression model is preferred. However, estimating the regression coefficients of the log-binomial model is challenging due to constraints on these coefficients. Bayesian methods provide a straightforward approach for log-binomial regression models, resulting in smaller mean squared errors and allowing for posterior inferences using WinBUGS software. However, the Markov chain Monte Carlo (MCMC) methods in WinBUGS can have a high Monte Carlo error. To address this issue, we propose an MCMC algorithm that uses a reparameterization based on a Poisson approximation and is designed to efficiently explore the constrained parameter space.
作者：Diego Salmer''on and Juan Antonio Cano
论文ID：1404.0042
分类：Computation
分类简称：stat.CO
提交时间：2014-04-02