Scholarlink

<p>In statistical analysis, the classical assumption of normal distribution of random observation errors in regression models is often violated, which covers up some important features in data. The main purpose of this article is to deal with the problem of non-normal hypothesis in measurement error model in which the dependent variable is censored. We extend the normal model by assuming that the latent covariate and the error terms have a finite mixture of scale mixtures of normal distributions. This approach can flexibly model data based on the structure of the mixture components to adapt to multimodality and heavy tails. We develop an MCMC algorithm to sample from the posterior distributions under the Bayesian paradigm. The analysis results of simulations and a real dataset demonstrate the effectiveness of the proposed method.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Bayesian analysis of censored measurement error models using finite mixture of heavy-tailed distributions

  • 期刊论文

  • 发表日期：

    2025年9月16日

查看全文
  • Jie Jiang,
  • Lichun Wang,
  • Liqun Wang

摘要

In statistical analysis, the classical assumption of normal distribution of random observation errors in regression models is often violated, which covers up some important features in data. The main purpose of this article is to deal with the problem of non-normal hypothesis in measurement error model in which the dependent variable is censored. We extend the normal model by assuming that the latent covariate and the error terms have a finite mixture of scale mixtures of normal distributions. This approach can flexibly model data based on the structure of the mixture components to adapt to multimodality and heavy tails. We develop an MCMC algorithm to sample from the posterior distributions under the Bayesian paradigm. The analysis results of simulations and a real dataset demonstrate the effectiveness of the proposed method.