部分线性反映变量含误差模型的经验似然置信区域

Empirical likelihood confidence regions for partially linear error-in-responses models

目的研究部分线性反映变量含误差模型。方法发展了经验似然比方法,将其应用到部分线性反映变量含误差模型。结果得到了W ilks定理的非参数形式,定理用来构造参数向量的渐近置信区域。结论参数β的置信区域也可由其他方法来构建,但与正态逼近或自助法相比,经验似然方法在实践中更直接和简单。

Aim To study partially linear error-in-responses models. Methods Develop the empirical likelihood method and apply it to partialy linear error-in-responses models. Results A nonparametric version of Wilks theorem is derived. The result is then used to construct asymptotical confidence regions of the parameter vector in the partially hnear error-in-responses models. Conclusion Confidence regions for β can also be constructed using other methods, but the empirical likelihood method is rather easier and more immediate in nractice.