ABSTRACT
It is suggested that past and current testing systems in education fail to provide pedagogically valuable information. The more potentially valuable approach to educational testing, criterion‐referenced measurement, is considered in detail. Although it should, theoretically, provide pedagogically useful information it is hampered by practical problems. These problems involve alternative interpretations of criterion‐referenced measurement; the setting of cutting scores; validity and reliability. Domain content and domain order determination also present difficulties.
It is concluded that the psychometric approach to criterion‐referenced measurement masks the quality of response by relying on the quantity of correct responses. The alternative is to redirect criterion‐referenced measurement principles to a learning theory base and a consideration of the quality of pupil errors.