“…Although examinations common to both PSI and lecture-discussion groups represent a more reliable basis, comparisons baaed on exams (4,9,16,18,23,24) and course grades have suffered from both the uncertain equivalence of the groups being compared due to the need to work with intact rather than randomly selected groups, and the differential loss of students. This threat to validity has been even greater where performance comparisons have been made between different student populations across different semesters on the basis of both examination scores (5,8,12,16,22) and course grades (2, 3, 6,8,26). Pasternack (19) addressed this problem by selecting PSI and lecture-discussion groups at random; however, interpretation of the results is complicated by the fact that not all students wrote the same exam, and scores were adjusted to account for differences in difficulty of the exams.…”