一种回归模型异方差性的检验方法

A Test for Heteroskedasticity in Regression Model

针对现有回归模型异方差性检验的局限性,借鉴尺度参数检验与重抽样方法的思想,通过对残差序列进行随机抽样,提出了一种可行的非参数检验方法;该方法具有更强的适用性,Monte Carlo模拟结果表明,其检验效果良好,尤其针对样本量较小的情形。

In view of the deficiencies of heteroskedasticity tests existing in regression model,by using the ideas of multi – sample scale parameter test and resampling method,through sampling the residual sequence randomly,a feasible non-parametric test method is proposed. The method has a stronger applicability,and Monte Carlo simulation shows that it has a good test power,especially in cases of small sample size.