Testing Problem Of Multivariate Balanced Random Effects Model

This article discusses the exact test problem and the problem of generalized confidence region of regression parameter matrix in multivariate balanced random effect model.In modern econometrics,random effect model is one of the most important components.With the development of econometrics,the random effect model has been developed.Therefore,the study of multivariate random effect model has practical significance and application value.On the one hand,it can improve econometric theory,on the other hand,it can also provide some theoretical support for empirical analysis.The first chapter is the introduction part,Firstly,research status of the random effect model and the test problem in multivariate model are introduced.Meanwhile,the research status of hypothesisn testing in linear model is briefly recommended.In addition,the study of generalized p-value is summarized.In the second chapter,the exect test problem and the problem of generalized confidence region of regression parameter matrix in multivariate balanced random effect model are investigated.In this paper,the method of generalized p-value is used to discuss an exact test of linear hypothesis of regression parameter matrix in multivariate balanced random effect model.Two independent submodels are obtained by transforming the multivariate balanced random effect model,and construct the two-step estimation of regression parameter matrix.Then,the generalized test variable is structured by using the two-step estimation,after that,the generalized p-value is defined.And through data simulation,compared with the method of likelihood ratio test,it is concluded that the test efficiency of the method given in this paper is better than that of likelihood ratio test.In addition,the generalized confidence region of regression parameter matrix is constructed by using the method of generalized pivot quantity,and simulate the coverage probability of the generalized confidence region of the parameter matrix by R software.The conclusions obtained in this paper are all new and feasible,the specific ideas are clear and definite,each problem is simulated with R sofeware,and the correctness and effectiveness of the conclusions are scientifically clarified.The results can enrich the research of multivariate random effect model in statistics and perfect econometrics theory.