| Causal mediated effects analysis is a method for exploring the relationship between treatment variables,mediating variables,and outcome variables.Its main purpose is to investigate whether and to what extent the treatment variables lead to the outcome through the mediator,and It is widely used in the fields of psychology,behavior,and statistics.In the existing studies,the mediators investigated by researchers are often scalar.This paper combines text analysis technology with causal mediation effect analysis,This paper aims to explore whether people’s emotions expressed in text can influence some macroeconomic indicators as an intermediary.Since the outcome variables are macro indicators collected on a daily basis,And people generate a lot of comments on online platforms every day,You can think of these people as a group,The emotional score of a group can be regarded as a distribution.Therefore,quantile vector is used in this paper instead of distribution intermediary.Most of the currently available causal mediator analyses deal with variables that are binary,and this paper gives the results when the treatment variables are continuous the two estimation methods for the causal effects and give the corresponding parameter identifiability assumptions.This paper verifies the validity of this paper’s method through simulation studies,and also explores that as the quantile vector dimension increases,the The change trend of the evaluation index of causal mediated effect estimation is explored as the dimensionality of the quantile vector increases.In addition,applying the proposed method to the actual data,the conclusions with practical significance were obtained. |
