Applications of decision theory to computer-based adaptive instructional systems

This paper considers applications of decision theory to the problem of instructional decision-making in computer-based adaptive instructional systems, using the Minnesota Adaptive Instructional System (MAIS) as an example. The first section indicates how the problem of selecting the appropriate amount of instruction in MAIS can be situated within the general framework of empirical Bayesian decision theory. The linear loss model and the classical test model are discussed in this context. The second section describes six characteristics essential in effective computerized adaptive instructional systems: (1) initial diagnosis and prescription; (2) sequential character of the instructional decision-making process; (3) appropriate amount of instruction for each student; (4) sequence of instruction; (5) instructional time control; and (6) advisement of learning need. It is shown that all but the sequence of instruction could be improved in MAIS with the extensions proposed. Several new lines of research arising from the application of psychometric theory to the decision component in MAIS are reviewed