Part of a larger project, "Technology, Educational Media & Materials--Assessing the Social and Cultural Impact of Group-Based, Videodisc Math Instruction in Mainstreamed Classrooms," this study addressed the question of whether the level of student achievement differs across levels of implementation of a mediated program that is based on a well-defined instructional model. Systematic observations of the implementation of a videodisc-based, direct instruction program, "Mastering Fractions," were conducted. The nine upper-elementary classrooms involved were spread across three school districts. Implementation/utilization of the program varied across classrooms, as did achievement. Based on the observations, three classrooms were classified as high-implementation classrooms, three as average-implementation, and three as low-implementation. At the conclusion of the field study, there was a 28-percentage-point difference between the mean gain score of the high-implementing classrooms and that of the low-implementing classrooms. When covariance-adjusted posttest means were considered, the difference was about 20 percentage points. Guidelines for teachers using the program, an implementation rating sheet, and a teacher interview questionnaire are appended. (20 references) (Author/BBM)