Errors-in-Variables Models

Abstract

Errors-in-variables (EIV) models axe regression models in which the regres-sors axe observed with errors. These models include the linear EIV models, the nonlinear EIV models, and the partially linear EIV models. Suppose that we want to investigate the relationship between the yield (Y) of corn and available nitrogen (X) in the soil. A common approach is to assume that Y depends upon X linearly. To evaluate the degree of dependence, it’s necessary to sample the soil of the experimental plot and to perform an analysis. We can not observe X, but rather an estimate of X. Therefore, we represent the observed nitrogen by W, also called the surrogate of X. The model thus studied is an errors-in-variables model.