Courtesy of Charles Gleek at Games Without Frontiers, I became aware of an interesting discussion of grade distributions at TPRS Q & A:
What grades should kids get? Notes on Evaluation for the Mathematically Challenged.
It’s a bit long but worth reading. The author, Chris Stolz, points out that perceptions about “proper” grade distributions are sometimes based on ignorance of basic statistical principles. Classes frequently are not statistically representative — they contain too few students, and students choose to enroll in them for non-random reasons. Assessment instruments often produce a ceiling effect that masks evidence of improvement in students who come into a course already possessing a high level of proficiency.
The end result can be a class composed of students who are either predisposed to do well in the course (possibly the main reason they enrolled in it to begin with) or who learn enough over the course of a semester to earn high marks on summative assessments. This reduces variation in the grade distribution and skews the curve to the right — instead of a normal distribution from F to A, with most students getting a C, the majority of the class ends up with A’s and B’s. A person who does not understand statistics assumes this happened because of grade inflation.
Several years ago I abandoned grading students against a normal distribution curve for these reasons. I also became much less concerned with testing students’ ability to reproduce factual information on “objective” exams because I knew that the vast majority of what they regurgitated would never move into their long-term memories. I did not (and still don’t) believe that their lives would be fundamentally altered for the worse if they failed to remember that the Turks captured Constantinople in 1453 or that realist IR theory derives in large part from the writings of Thucydides, Hobbes, and Machiavelli.
I thought that students would benefit more from multiple opportunities to demonstrate how well they could apply concepts in novel ways and effectively communicate their findings. How does this look in reality? Below is my grading system for a course that I’m teaching now.
In this course, final letter grades are based on a 1,000 point scale, in which students need only earn 950 points to obtain an A. Obviously with a total of 1,080 points available, it’s quite possible for a student to earn a high grade if he or she simply keeps plugging away at all the various assignments. But this is exactly what I want — for many students, continuous effort will result in improvement across the semester. Constant practice is also makes it more likely that students retain something after the course ends. And students feel better about themselves and their environment with frequent feedback on their performance.
Since this system of assessment makes it more likely that students will be able to demonstrate proficiency by the end of the semester, my grade distribution shifts to the right. Is this grade inflation? I will argue that it isn’t, because the student’s final grade is not based on a hastily thrown together end-of-semester essay that the instructor simply marks as an A or B.