Choosing Minimum Passing Scores By Stochastic Approximation Techniques

Abstract

Often a written test is used as an inexpensive substitute for a per formance measure. A specified minimum performance level or prob ability of successful performance can be translated into a minimum passing score for the written test most efficiently by measuring the performance of students whose written test scores are near the de sired cutoff score. Stochastic approximation methods accomplish this purpose. The up-and-down method and the Robbins-Monro process are presented, discussed, and compared.