Student Assessment Precision in Mechanical Engineering Courses

Student performance on mid-term quizzes and final examinations was compared for several core undergraduate courses. A statistical comparison of the grades of students on these two tests showed only a modest correlation between grades. The modest correlation may be attributed to several factors: inaccurate grading of examinations, differences in the material tested on the mid-terms and finals, and sampling error due to the small numbers of questions on examinations. Evidence is presented that sampling error is the primary source of the low correlations. In practice, the modest correlations of mid-term quiz and final examination grades imply that the grades assigned to students are some aggregate of a true measure of the ability of students and randomness. The randomness arises because the instructor selects a particular subset of the course material for examination, not complete coverage of the course by question of all levels of difficulty.