Panel Data模型基于Within估计的最优设计  

Optimal Designs of Panel Data Model Based on Within Estimator

作　　者：程靖 岳荣先[2] CHENG Jing;YUE RongXian(College of Science,Anhui Agricultural University,Hefei,230036,China;College of Mathematics and Sciences,Shanghai Normal University,Shanghai,200234,China)

机构地区：[1]安徽农业大学理学院统计系,合肥230036 [2]上海师范大学数理学院,上海200234

出　　处：《应用概率统计》2018年第1期95-100,共6页Chinese Journal of Applied Probability and Statistics

基　　金：国家自然科学基金项目(批准号:11401056;11471216);安徽省环保厅项目(批准号:2015-5);安徽农业大学稳定与引进人才项目(批准号:yj2015-27)资助

摘　　要：本文研究二元Panel Data模型基于Within估计在恒等设计类中的D-,A-,I-和E-最优设计.文章首先证明了二元Panel Data模型基于Within估计在恒等设计类中的几类最优设计可在设计域的顶点处获得;研究发现模型的D-,A-和E-最优设计均为设计域顶点上的等权重设计,I-最优设计为对称设计;结果还表明等权重设计具有较高的I-效.In this paper,optimal identical designs of bivariate panel data model based on within estimator are discussed.It is proved that D-,A-,E-and I-optimal design of bivariate panel data model can be obtained on the vertexes of the design region.Equal-weight design is proved to be D-,A-,E-optimal of bivariate panel data model in the paper.I-optimal design is also obtained in the paper and it shows that equal-weight design has high I-efficiency.

关 键 词：PANELDATA模型 Within估计 最优设计 

分 类 号：O212.6[理学—概率论与数理统计]