Properties of a consistent estimation procedure in ultrastructural model when reliability ratio is known

Abstract

To overcome the problem of inconsistency of the ordinary least squares (OLS) estimators for regression coefficients in a linear measurement error model, we assume the reliability ratio to be known a priori and utilize this information to form consistent estimators in an ultrastructural models. Assuming the distributions of measurement errors and the random error component to be not necessarily normal, the efficiency properties of the estimator for slope parameter have been analysed and the effect of departure from normality is examined.