This discussion is concerned with what has been learned through systematic research and accumulated practical experience about how to measure the outcomes of formal instruction in schools and colleges. According to Adkins (1958) we may think of achievement as consisting of (a) increases in the variety of stimulus dimensions to which the learner will be sensitive and responsive, (b) increases in the number of new responses that will be made in the presence of familiar, or already discriminated, stimulus components, or (c) increases in the number of new responses to be made in novel stimulus situations. The emphasis on change which Adkins makes calls attention to the fact that, although an achievement test can provide only a measure of status at a particular time, interpretation of achievement test scores in relation to educational programs must of necessity involve, either explicitly or implicitly, the collection of data at two points in time. If there is any single limitation which more than any other characterizes the evaluation practices of educators at all stages of the educational system it is the tendency to accept as evidence of achievement the performance of students at a single point in time without regard to differences in achievement at some defined previous point in time. The result is that schools are credited with producing high quality output when the output depends primarily on high quality input, and pupils are charged with deficiencies in effort when they are actually achieving at normal levels in relation to their abilities and backgrounds.