What Should Grades Mean?

Michael A. Covington
Institute for Artificial Intelligence
The University of Georgia
http://www.ai.uga.edu/mc/grading.html

Originally published as "A Philosophy of Grading" in Teaching at UGa, Spring 1989. [Updates added in 2004 are in brackets.]


Both students and teachers benefit from a grading policy that is explicit, well-thought-out, and designed to head off disputes. Over the years Iíve been developing such a policy, and this essay describes it. Iím writing to defend current practice, not reform it, so for the most part Iíll be giving reasons why weíre justified in doing what we already do.

My first premise is that grades are a system of communication and are worthless unless everyone uses the the same way. We issue transcripts so that employers and graduate schools will know how well our students have performed. Hereís how people across the country will interpret the grades we give:

A The student did as well as could reasonably be expected.
B The studentís mastery of the material has noticeable flaws but is well above the minimum standard.
C The student met the minimum requirements for the course.
D The student learned some of the material but did not meet minimum requirements.
F The student learned little or none of the material.

Thereís no point in saying the letters "ought" to mean something else - for instance that C "ought" to mean "average" today, as it did decades ago. When you speak a language, you have to use it the way other people use it, or they wonít understand you.

Second, grades measure results, not effort. It sounds charitable to "grade on effort," rewarding hard work with higher grades even in the absence of measurable results. But this wonít work in practice. We have no way to measure how hard our students work. When you give one failing student a C for effort, are you sure you didnít have another failing student who worked just as hard but didnít tell you about it?

Anyhow, we give degrees for mastering course material, not for enduring drudgery, and a grade is part of a degree. Would you rather be operated on by a doctor who had an easy time in medical school, or by one who had a hard time and got his passing grade "on effort"? Grading on effort can conceal incompetence or, at best, send students into advanced courses for which they are not prepared.

Third, grades are not numbers. We use numbers to ensure that grades are given fairly, but the grade itself is a teacherís professional judgment of a studentís knowledge. It does no good to tell the world that a student scored between 85 and 90 on your exam, unless you also tell the world exactly what is on the exam. Likewise, it makes little sense to issue arbitrary declarations that "B is 85 to 95." The University publishes course descriptions, not exams.

Fourth, students should be graded against course requirements, not against other students. A certain amount of competition is healthy, but a student should not suffer because his fellow students do well, nor benefit because his fellow students do badly. A graduate school or employer will want to know whether this student mastered the contents of the course, not whether his fellow students failed to do so.

Non-statisticians often donít realize that "grading on a curve" works well only when the class is enormous - with perhaps 1000 students - and the student performance is known to be constant over time. In a class of 20 or 50, random variation will invalidate any attempt to give Aís, Bís, and Cís to fixed percentages of the class; some students will be sure to get different grades than they would have gotten for the same work in a different quarter [or semester].

In any case, performance can change from year to year, and grading on a curve allows standards to sink to any depth as long as all students sink together.

Fifth, lowering grades is no substitute for beefing up content. "Grade inflation" is a proper concern; the average grades in some of our introductory courses have become too high to be informative [in 1988, when this was written]. But the way to counteract this is to require students to learn more material - not to lower their grades by mathematical trickery. (A related problem is that instructors sometimes give an excessive workload to counteract claims that a subject is "easy." Workload is not the same thing as content.)

In advanced courses, it can be quite proper for all students to get Aís and Bís, because weak students would not take the course in the first place. The all-A graduate seminar is perfectly legitimate. Conversely, if no one ever gets an A in a course, the course should be changed; it isnít meeting the needs of its students.

Finally, low grades are not a punishment for misconduct. Some instructors give Fís as the only punishment for students who are caught cheating. I object to this for two reasons. First, it is no deterrent to students who are cheating to avoid an F - the usual case. Second, and more seriously, it exposes the teacher to legal liability if the accusation is mistaken.

I refer all cases of suspected cheating to the Office of Judicial Programs. [Procedures have changed slightly since 1988.] Although this sounds heavy-handed, itís the only way to ensure that the student gets a confidential investigation by experts followed by a fair hearing. Afterward, the student canít sue me or claim I was out to get him. Of course, if itís proven that he was indeed dishonest, he still gets the F that he earned.




The content and opinions expressed on this Web page do not necessarily reflect the views of,
nor are they endorsed by, the University of Georgia or the University System of Georgia.