Causal Inference Based On Multiple Imputation

In all ages,many activities in production and life are based on causal inference.Exploring causal relationships between complex things is an important part of research in many fields such as epidemiology,economics,statistics and social sciences.Therefore,the study of causal inference has extremely important theoretical and practical significance.In the research about causal inference,an important causal model is the counterfactual model,also known as potential outcomes model.However,since observations from a single individual can only yield one result,the introduction of potential outcomes leads to in a lot of missing in the data set.This article mainly focuses on the research of missing data in the counterfactual model in causal inference.The method of multiple imputation is used to fill in the missing data in the counterfactual model and further estimate the causal effect.In this paper,we first introduce the relevant knowledge of the counterfactual model,and give multiple imputation when the potential outcomes are missing.This method fills in missing information of potential outcomes in counterfactual model and estimates the average causal effect based on the complete data set.In addition,we have improved the log-binomial model based on multiple imputation method.Interpolation of missing values for missing potential outcomes improves the estimation accuracy of the average causal effect.Finally,the multiple interpolation method is applied to the missing data of 80084 traffic accidents to get a better result for estimating average causal effect.