Abstract
To find the bias and mean square error of ratio estimator, it is usually assumed that for all possible samples. Koop [2] suggested a new approach to overcome this assumption. This note examines Koop's approach and shows that the assumption is still necessary.