上海财经大学 > 科学研究 > 学术交流 > 学术报告
  • 统计与管理学院2017年学术报告第6

     

    【主  题】Bayesian Empirical Likelihood Estimation of Quantile Structural Equation Models

    【报告人】 唐年胜 教授

    云南大学

    【时  间】 2017年03月17日(星期五)14:30-15:30

    【地  点】 上海财经大学统计与管理学院大楼1208室

    【摘  要】Structural equation models (SEMs) are often proposed to investigate the interrelationships among latent variables in biomedical, educational, behavioral, psychological, medical and social sciences. In traditional SEMs,  random errors are assumed to follow a family of normal distributions, and a mean regression-type structural equation is posited to assess the effect of explanatory latent variables on manifest variables. However, these SEMs may be unreasonable in the cases with highly nonnormal distributed errors, and cannot provide a comprehensive analysis of the relationships among latent variables. This paper incorporates estimating equation (EE) approach and quantile regression method into a SEM to investigate the performance of the exogenous latent variables given the endogenous latent variables and covariates under different quantiles. A Bayesian empirical likelihood method together with Markov Chain Monte Carlo (MCMC) algorithm is proposed to simultaneously estimate model parameters and latent variables. Latent variables are regarded as missing data, and imputed by the estimated density function and the linear interpolation method. A simulation study is conducted to investigate the performance of the proposed methods, and a real example is illustrated.

    【邀请人】 周勇