A Comparison of Branching Methods for Programmed Learning

Abstract

IN EARLIER STUDIES with linear programs, where each student had to go through all the items in a predetermined sequence, we failed to find any significant differences in the terminal performance of students who had worked with multiple choice machines, composed response machines, pro grammed texts requiring overt responses, and programmed texts requiring no responses. How ever, significant differences had been obtained in the amount of time required to complete the learn ing session. We therefore decided to investigate those methods of presenting programmed material which might speed up the learning process without adversely affecting terminal performance. Many varieties of branching programs have been suggested to accomplish this purpose ( backward branching to missed items, backward branching to review an entire sequence of items, backward branching to alternate form items, lateral branch ing to supplemental or prerequisite material, branching down to more detailed items for slow students, branching up to faster programs for bright students, forward branching, etc.), but very few experiments^ have compared the relative merits of the various branching methods. Early in the analysis of the problem, we concluded that all types of branching and linear programs are re ducible to one common prototype form (selective forward branching), and that the distinction be tween the various name-type s is more arbitrary than is commonly suspected . Furthermore, by casting all of the different types of branching procedures we wished to investigate into the pro totype form, we could physically present all branching formats in one simple inexpensive device.