A Compromised Estimator in the Measurement Error Model with Error Variances Unknown

Abstract

We consider the linear measurement-error model in which both the dependent variable, y, and the independent variable, x, are subject to measurement errors. We propose a compromised estimator of the slope paramater. The estimated line lies between the two extremes: the regression line of y on x and that of x on y With normally distributed errors, we show that this estimator has lower moments. The approximate expectation, variance and mean spuared error are obtained by Taylor series expansion.