tested against empirical data. The effect of number of items by number of points on the
sampling error of a composite scale value estimator is investigated and quantitatively
formulated. An application of this formulation to arrive at optimal combinations of number of
points and number of items is described.