Two studies attempted to determine whether a programed instruction booklet on a task in algebra would be more effective if the units in the program were presented in a sequence suggested by a hierarchical analysis of the task than if units were presented in a reverse or a random sequence. In the first study, a program was made up, based on the Princeton Algebra Program, in which blocks of frames made up units corresponding to the competencies in the hierarchy developed by Gagne and Paradise. Three versions of the program--forward order, reverse order, random order--were used with groups of eighth-graders in a school with a large proportion of disadvantaged students. The results yielded many trends in differences among the three versions of the program, but few were statistically significant. The second study was a replication using older students from more advantaged background and only the forward and random order programs. The forward version of the program resulted in better performance for the older and more advantaged children than for the younger ones in the first study. The highest retention scores however came from younger, less advantaged subjects who had the random sequence in the first study. These results and other significant findings are discussed in some detail. (JY)