Comment for identification and estimation of nonlinear models using two samples with nonclassical measurement errors, by Carroll, Chen and Hu

Abstract

This is a very interesting paper that develops nonparametric identification results and and semiparametric estimators for a nonparametric and semiparametric nonclassical measurement error model using a combination of a primary data set and an auxiliary data set. Their estimator not only achieves the semiparametric efficiency bound when the conditional regression model is correctly specified parametrically, but also performs well in finite sample simulation designs. In their paper, an application of their method to studying the relation between the amount of beta-carotene from food and the latent true daily long-term intake of beta-carotene using two data sets from the Eating at America's Table Study (EATS) and the Observing Protein and Energy Nutrition study shows that ignoring measurement errors in the EATS data set leads to substantial attenuation bias in the regression coefficient.