Effect Estimations Of Causal Mediation Analysis

Causal mediation analysis is to investigate the effect of independent variables on dependent variables through intermediate variables,and its purpose is to reveal the intermediate relationship between independent variables and dependent variables.Therefore,the mediation effect is a multiple-causality relationship consisting of independent variables,mediation variables,and dependent variables,and its essence is a chain of cause-effect relationships.So far,the research results of causal mediation analysis have mainly focused on the case of a mediation variable.However,there may be more than one intermediary variable that affects the dependent variable.Therefore,in order to solve the problem that there are multiple mediating variables,this paper studies a mediating model with multiple mediating variables and a mediating model with multiple mediating variables with sensitivity.The work of this thesis mainly includes the following two parts.The first part is mainly aimed at the problem that there are multiple mediating variables.Based on a simple mediating model,combining the interaction terms of independent variables and mediating variables on the dependent variables,a mediating model with multiple mediating variables is proposed,and The direct and indirect concepts under multiple intermediary variables are defined,and the total effect is decomposed into direct and indirect effects.The intermediary formula with multiple intermediary variables is constructed,and the estimation of intermediary effects is proposed.Finally,a simulation study of the proposed estimates is performed,and the simulation results show that the proposed method performs wellThe second part is based on the simple mediation model,and proposes a mediation analysis model with multiple mediation variables with sensitivity parameters and a mediation analysis model with multiple mediation variables with interactions of sensitivity parameters.First,for the setting of sensitivity parameters in the intermediary model,different parameter settings are used.Secondly,the binomial distribution is adopted for the independent variable setting.Finally,the mediating formula of multiple mediating variables is given in combination with the sensitivity parameter problem,the estimation of mediating effect is simulated,and the simulation results are analyzed.The simulation results show that the proposed method performs well.Figure[9]Table[10]Reference[58]...