Assessing Student Work

But in 1912 a study by two Wisconsin researchers seriously challenged the reliability of percentage grades as accurate indicators of students' achievement.

These demonstrations of wide variation in grading practices led to a gradual move away from percentage scores to scales that had fewer and larger categories. One was a three-point scale that employed the categories of Excellent, Average, and Poor. Another was the familiar five-point scale of Excellent, Good, Average, Poor, and Failing, (or A, B, C, D, and F). This reduction in the number of score categories served to reduce the variation in grades, but it did not solve the problem of teacher subjectivity.

At the same time, significant evidence shows that regularly checking on students' learning progress is an essential aspect of successful teaching–but checking is different from grading. Checking implies finding out how students are doing, what they have learned well, what problems or difficulties they might be experiencing, and what corrective measures may be necessary. The process is primarily a diagnostic and prescriptive interaction between teachers and students. Grading and reporting, however, typically involve judgment of the adequacy of students' performance at a particular point in time. As such, it is primarily evaluative and descriptive.

To ensure a fairer distribution of grades among teachers and to bring into check the subjective nature of scoring, the idea of grading based on the normal probability, bell-shaped curve became increasingly popular. By this method, students were simply rank-ordered according to some measure of their performance or proficiency. A top percentage was then assigned a grade of A, the next percentage a grade of B, and so on. Some advocates of this method even specified the precise percentages of students that should be assigned each grade, such as the 6-22-44-22-6 system.

Grading on the curve was considered appropriate at that time because it was well known that the distribution of students' intelligence test scores approximated a normal probability curve. Since innate intelligence and school achievement were thought to be directly related, such a procedure seemed both fair and equitable. Grading on the curve also relieved teachers of the difficult task of having to identify specific learning criteria. Fortunately, most educators of the early twenty-first century have a better understanding of the flawed premises behind this practice and of its many negative consequences.