This final report of a two-stage project describes an effort to determine whether students receiving instruction in freshman English composition perform better on standardized tests than students who do not receive similar instruction, when both groups are in college the same length of time. The second phase of the experiment detailed in the report involves 1,040 matched pairs of students from the University of Northern Iowa, the University of Iowa, Kent State University, the University of Colorado, and Northern Illinois University. Using the Cooperative English Tests: English Expression (COOP), the College Entrance Examination Board (CEEB) English Composition Test, and a theme as test instruments, the authors include computer-generated test results covering frequent intervals over the 2-year period on: (1) overall performance, (2) performance by ability quarters by sex, and (3) performance by sex. Background, procedures, summary, and recommendations are included. Forty-eight statistical tables reveal performance data. (RL)