经验似然重抽样下回归模型的Bootstrap逼近(英文)

Bootstrapping Regression Models via Empirical Likelihood Resampling

本文提出用经验似然重抽样来bootstrap逼近线性回归模型中的学生化最小二乘估计．我们证明了该方法具有一般s－2项Edgeworth展开，它是二阶相合的而且比经典的方法损失更小．

In this paper an empirical likelihood resampling is proposed for bootstrapping studentizedleast square estimation in linear regression models. It is proved that our method captures ageneral s-2 term Edgeworth expansion and achieves a second order accuracy, furthermore, ithas smaller loss than the classical one in most cases.