OPTIMAL PROCEDURE FOR AN N-STATE TESTING AND LEARNING PROGRESS.Ⅱ

摘要

This paper is a continuation of our work on optimal strategies for presenta¬tion of items in an N-trial learning experiment. In SP-1869 and SP-1922 we showed (under certain assumptions) that the following decision rule generated an optimal sequencing: in any trial present the item for which the probability of being in the learned state is least. In the present paper we show that this rule is optimal for a more general learning model than any considered earlier;the new model allows for the possibility that a subject may respond incorrectly . to a test on an item, even though he "knows" the item.