One component of an evaluation research model for system-wide selection of curriculum materials is reported: implementation of an evaluation design for obtaining data that permits professional and lay persons to base curriculum materials decisions on a "best fit" principle. The design includes teacher characteristics, learning environment variables, and instructional variables, in addition to materials content analysis and achievement data. After removing the effects of pretest achievement, students' perception of competitiveness in the reading class was a significant cross-validated variable affecting posttest achievement. The findings were instrumental in articulating the system's philosophy of reading and in selecting the "best fitting" reading series. (Author/RC)