The Use of Uniformative and Informative Prior Distribution in Bayesian SEM

Yanuar, Ferra The Use of Uniformative and Informative Prior Distribution in Bayesian SEM. - .

PDF (Author’s self archive) Download (115Kb)

Abstract

This study applies Bayesian approach to the construction of health status model. Bayesian combines prior distributions with the data likelihood to form posterior distributions to estimate the parameters. An algorithm based on the Gibbs sampler is applied for drawing the parameters values from the joint posterior distributions. Some criteria are used to test the goodness of fit of the posited model. Uninformative and informative prior are used in Bayesian analyses. Since health status model involves observed and unobserved variables simultaneously, Bayesian analysis is then combined with structural equation modeling (SEM) approach in fitting the hypothesis model to the data. The main purpose of this study is to demonstrate the application of uninformative and informative prior in Bayesian SEM to construct the health status model of an individual. Two real data sets are considered in this study. First, data set uses uninformative prior in parameter estimation, which then be adopted as informative prior for the second data set. This study proves that model uses informative prior results better estimation on parameter model than uninformative. This study also informs that socio-demography and lifestyle have greater effect to the health condition of an individual than to mental health.

Actions (login required)

View Item