Document Type



In 2 experiments, each involving different mathematical operations, we compared 2 training procedures for teaching component math skills in terms of their effects on the learning and long-term maintenance of composite skills. The dependent variables were learn units to composite task mastery and performance on the composite task 2 months later. The independent variables were instruction in math facts under (a) fluency and (b) mastery conditions. The experiments used a simultaneous treatment design in which the students were selected for participation according to prerequisite skills and instructional histories and randomly assigned to receive 1 of the 2 training procedures. Four adolescents with developmental disabilities participated in each experiment. Instructional presentations were controlled by yoked learn units during component skill instruction. Results showed that fluency instruction did not result in fewer learn units to criterion on the composite task. However, 2 months later, the fluent students performed between 83% and 100% correct on the composite task, while the mastery students performed between 17% and 83% correct. The data are discussed in terms of fluency theory and educational practice.