The Effect of Sequence on Programmed Instruction1

Of all the criteria of a good linear program, the one which seems hardiest is that of sequence. Other criteria such as overt response, small steps, and response mode have all been questioned successfully, and so, for that matter, has sequence. But sequence continues to be widelyaccepted as a criterion on an intuitive basis, even if it is not always supported empirically. No one seems to doubt, that were one to scramble a whole course that learning would be retarded, so that in part, the size of the unit in which sequence is destroyed is a factor. But as Levin and Baker (1963) point out, perhaps equally critical is the apparent logical dependence upon sequence within the material to be learned. Thus there may be a continuum of dependence on sequence. At one extreme of the continuum, scrambling may have no effect on learning a set of spelling words which has no logical structure. At the other extreme, scrambling would be expected to result in considerable decrease in learning if the learning of one concept were prerequisite to learning the next in a logical hierarchy. In previous studies of sequencing the researchers used programs which were supposed to be at the latter extreme of the continuum, but the position of the programs on the continuum, but the position of the programs on the continuum was not systematically examined as an independent variable. On learning elementary probability from a 71 frame multiple choice program, Gavurin and Donahue (1961) scrambled 29 items of the Holland and Skinner (1961) program within approximately 10 item blocks and required an errorless trial within a block before the subject proceeded to the next block. Although the scrambled program group took more trials to criterion, retention a month later showed no statistically significant difference in comparison with conventional use of the program in proper sequence.