Statistics fail to solve grading contrasts

Students in my fall course generally get the same (high) grade, though they and the quality of their work differ considerably, one from the other. If you infer from this-as the proponents of Achievement Index evidently do-that I lack standards and give "soft" grades, consider this: The grade in my course depends upon the quality of a research paper that is based upon an original idea that leads to an appropriate research design and data collection.

One student has a stunningly original, important idea and develops the appropriate methodology with which to test it, but her paper is inconclusive (and unpublishable) because she lacked time to collect sufficient data with which to test her ideas. Another has a more humdrum idea, but his design is clever and the data he generates clearly allows the rejection of some but not all other hypotheses. He writes well, too, so this paper will likely appear in print.

A third student has a good idea, develops the experimental design cleverly, collects adequate data but, alas, lacks the literary skills to convincingly present the results. With time and assistance, this paper will assume publishable form, too, but not before the term's end. And so on...

Identical grades in these instances, I suggest, are an inevitable result of our trying to describe different capacities with a single score. When I write letters of recommendation for these three students, I can distinguish between their strengths and weaknesses; when I report their performance to the registrar, those differences are collapsed into a single grade.

This is a problem that no degree of post hoc adjustment by our statistician friends can remedy. Perhaps stat courses, where test questions have but one possible answer, permit a meaningful ranking of students. I suggest that courses with intellectual content rarely allow anything so simplistic. Why worry about it? Grades, except when applied to meat, are so crude a measure of ability and accomplishment, that no reasonable person should take them at face value. Further discussions on grading policies are, I think, best left for those who enjoy trivial pursuits.

Peter Klopfer

Professor

Department of Zoology