蒙特卡罗方法下线性模型的异方差性检验方法  

Heteroscedasticity Testing Method by Linear Model Under the Monte Carlo Method

作　　者：王佐仁[1,2] 徐生霞 

机构地区：[1]西安财经学院统计学院,陕西西安710061 [2]西安财经学院西安统计研究院,陕西西安710061

出　　处：《统计与信息论坛》2016年第11期33-37,共5页Journal of Statistics and Information

摘　　要：在已有的异方差性检验方法的基础上,运用蒙特卡罗方法,借助permutation检验思想,在不假定随机扰动项服从同一分布族的条件下,通过从大样本中提取大量的子样本,不断对线性模型进行拟合和检验,根据异方差为真的频率大小,给出了一种新的异方差检验方法。随机模拟表明本检验方法优于传统方法。Existing Heteroscedasticity testing method is presented in this paper,on the basis of using the Monte Carlo method,with the help of a thought of permutation test,without assuming that submitting to the identically distribution family of the random disturbed term,through a large amount of extracted from large sample sample,constantly to fit linear models and inspection,according to Heteroscedasticity for really frequency size,put forward a new Heteroscedasticity testing method is given.Stochastic simulation shows that the test method is better than traditional methods.

关 键 词：异方差性 蒙特卡罗 随机模拟 统计检验 

分 类 号：F224.0[经济管理—国民经济] O212[理学—概率论与数理统计]