摘要
小域估计问题日益受到社会各界的关注,它通常利用辅助信息和统计模型提高估计的精度。其中最常用的小域模型是混合模型,即利用域随机效应来解释域间变化,但是这种模型要求严格的假定条件,不易于处理实际中存在异常值或重尾现象的小域估计问题。本文将分位数回归模型引入小域估计中,这个模型不需要强的假定条件,可以处理实际中存在异常值或是重尾现象的小域估计问题,并通过一个模拟案例进一步说明了基于分位数回归模型的小域估计方法可以得到更加稳健的估计量,挖掘更多的信息来提高小域估计的精度,是一种比较好的小域估计方法。
Small area estimation gets more and more attention. It generally makes use of the auxiliary information and model to get effective estimators. Linear mixed model is a common model which containing random area effect to characterize area variability, but this model needs strong suppositions and cannot deal with practical data. In this paper, we apply the quantile regression model to small area estimation which doesn't need strong suppositions and can deal with practical data, besides, we use a simulation case to illustrate that it is a good method to do small area estimation based on quantile regression model, it can get robust estimators and mine more information to improve estimators.
出处
《统计教育》
2009年第1期56-59,64,共5页
Statistical education
关键词
分位数回归模型
小域估计
混合模型
稳健估计量
Quantile regression model
Small area estimation
Mixed model
Robust estimator