The Research On Causal Effect With Latent Variables

Cause and effect is a basic relationship between the complex things.Not only in real life there are a lot of causal relationship,but also it have been follow with scientific research workers in statistics,medicine,economics and many other fields and it has an important application.The introduction of latent variables is very important in the study of cause and effect.It not only clearly and profoundly depicts the relationship between variables,but also makes a great contribution to simplify complex problems.For the study on the causal effect of latent variable model,this paper does it from three aspects.Firstly,it is the virtual reality model.Secondly,it is the causal network model,and thirdly it is based on factor analysis to explore the causal relationship.The first part is the study of virtual reality model framework.We compared the effects of VanderWeele(2008)and Chiba(2009)on average causal effects.We can come to this conclusion by practical examples and simulation studies.In most cases,Chiba's boundary compared with VanderWeele and the former method is better,because his method can get a smaller boundary,and this small boundary can make our study more accurate,and it can also get more accurate results in practical applications.The virtual reality model is to evaluate the causal effect of one variable on another by introducing virtual reality variables,but the causal network model represents the causal relationship between variables through network structure table,and it can be used to judge whether the causal relationship between variables exist(the edge of a network)with the help of graph model in the presence of latent variables.This article explores the causal relationship between variables such as hospital time,physical condition,and survival status through NSCOT data.We can conclude that there is a causal relationship between the hospital time and the living conditions,and a person's physical quality and the severity of the injury will affect the patient's time to the hospital,thus affecting the living conditions.The third research question is factor analysis.Factor analysis is a statistical method,the main idea is to use a few factors to describe the relationship between a number of indicators,and then a few factors to reflect the majority of information in the original data.In this paper,we use the quality of life data of different provinces,and we can get three common factors,which are the overall factors,individual factors and consumer factors.These three factors are latent variables which can't be directly observed.