Key points are not available for this paper at this time.
A computer program that presents and grades physics problems is described. The reliability of the grading is discussed in the context of diagnostic tests. The concept of computerized testing in physics is shown to be feasible. It was found that graphics answer representations can be used to measure student understanding: students with correct graphics answers provide correct explanations and conversely. The sample problem studied did not produce any false negative results; the only false positive results were caused by incomplete explanations which are easily caught by keyword checking.
Lea et al. (Sat,) studied this question.