Empirical Likelihood Confidence Regions in QVF Measurement Error Models
In applications in many areas, the data sets are contaminated or corrupted by the mismeasured covariate variables resulting in the so-called measurement errors. In this article, we propose an empirical likelihood method for constructing confidence regions in measurement error models. The quasi variance functions are used to establish functional constraints for the empirical likelihood method. The proposed empirical likelihood confidence regions in measurement error models as a nonparametric approach has clear advantages over the parametric model approaches. The distribution of the response error as well as the distribution of the error prone covariates does not need to be specified. Simulation studies in a simple linear model and a logit regression model are conducted. The results show that the proposed method performs well.
Shifa, Naima; Chen, Hanfeng; and Wu, Yuehua, "Empirical Likelihood Confidence Regions in QVF Measurement Error Models" (2015). Mathematics and Statistics Faculty Publications. 35.
Journal of Probability and Statistical Science